merge: resolve conflict with litellm_internal_staging in anthropic transformation tests

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

.circleci/config.yml

@@ -1483,7 +1483,7 @@ jobs:
       - run:
           name: Run tests
           command: |
-            uv run --no-sync python -m pytest -vv tests/local_testing/test_basic_python_version.py -k "not v2_resolver"
+            uv run --no-sync python -m pytest -vv tests/local_testing/test_basic_python_version.py -k "not legacy_resolver"
 
   installing_litellm_on_python_3_13:
     docker:
@@ -1507,9 +1507,9 @@ jobs:
       - run:
           name: Run tests
           command: |
-            uv run --no-sync python -m pytest -v tests/local_testing/test_basic_python_version.py -k "not v2_resolver"
+            uv run --no-sync python -m pytest -v tests/local_testing/test_basic_python_version.py -k "not legacy_resolver"
 
-  installing_litellm_on_python_v2_migration_resolver:
+  installing_litellm_on_python_legacy_migration_resolver:
     docker:
       - *python312_image
       - image: cimg/postgres:16.0@sha256:b125148bc76e8e8eee5eb3ad6020a3a14110a14e8192f1c645128afebe2e2f84
@@ -1536,10 +1536,10 @@ jobs:
           url: tcp://localhost:5432
           timeout: "60"
       - run:
-          name: Run v2 migration resolver proxy smoke test
+          name: Run legacy migration resolver proxy smoke test
           command: |
             uv run --no-sync python -m pytest -vv \
-              tests/local_testing/test_basic_python_version.py::test_litellm_proxy_server_config_no_general_settings_v2_resolver
+              tests/local_testing/test_basic_python_version.py::test_litellm_proxy_server_config_no_general_settings_legacy_resolver
 
   helm_chart_testing:
     machine:
@@ -2879,7 +2879,8 @@ jobs:
           command: |
             if grep -q "Error: P1001: Can't reach database server at" docker_output.log && \
                (grep -q "Database setup failed after multiple retries" docker_output.log || \
-                grep -q "ERROR:    Application startup failed. Exiting." docker_output.log); then
+                grep -q "ERROR:    Application startup failed. Exiting." docker_output.log || \
+                grep -q "Database migration cannot proceed" docker_output.log); then
               echo "Expected error found. Test passed."
             else
               echo "Expected error not found. Test failed."
@@ -3011,7 +3012,7 @@ workflows:
           filters: *main_branches
       - installing_litellm_on_python_3_13:
           filters: *main_branches
-      - installing_litellm_on_python_v2_migration_resolver:
+      - installing_litellm_on_python_legacy_migration_resolver:
           filters: *main_branches
       - helm_chart_testing:
           requires:

CLAUDE.md

@@ -37,7 +37,7 @@ Same applies for filing bug reports and feature requests, with .github/ISSUE_TEM
 
 If you're resolving a linear ticket, in the "## Linear ticket" section of the PR, say "Resolves LIT-1234", replacing "LIT-1234" with the actual ticket id that you're resolving. If you don't have the ticket id, don't make one up or search for it. Just leave the section blank
 
-Never use `pytest` commands or the like as "Screenshots / Proof of Fix". We prefer curl'ing a live proxy instance running on localhost:4000 (I like to run it with `python litellm/proxy/proxy_cli.py --config litellm/proxy/dev_config.yaml --detailed_debug --reload --use_v2_migration_resolver 2>&1 | tee litellm.log`; the Admin UI dev server is `npm run dev` in `ui/litellm-dashboard`, served on port 3000) and showing both the command run and the output. Also, it should hit real LLM provider APIs, not mocks, and cost real $$$ because that is the most realistic test. The proof of fix should be exactly what the end user / customer would see / do. The run logs in PR #27703 is a prime example of how to do it (not a huge fan of using a python test script that future me and the team will have no visibility into; I prefer just curl commands or a short list of bash commands (e.g., using `for`)). If it's a UI thing, just tell me which URLs to go to (e.g., http://localhost:4000/ui/?page=logs), where to click, what fields to fill out, etc. along with the other commands to run in an ordered list, and I'll do it myself and post the screenshots after you make the PR
+Never use `pytest` commands or the like as "Screenshots / Proof of Fix". We prefer curl'ing a live proxy instance running on localhost:4000 (I like to run it with `python litellm/proxy/proxy_cli.py --config litellm/proxy/dev_config.yaml --detailed_debug --reload 2>&1 | tee litellm.log`; the Admin UI dev server is `npm run dev` in `ui/litellm-dashboard`, served on port 3000) and showing both the command run and the output. Also, it should hit real LLM provider APIs, not mocks, and cost real $$$ because that is the most realistic test. The proof of fix should be exactly what the end user / customer would see / do. The run logs in PR #27703 is a prime example of how to do it (not a huge fan of using a python test script that future me and the team will have no visibility into; I prefer just curl commands or a short list of bash commands (e.g., using `for`)). If it's a UI thing, just tell me which URLs to go to (e.g., http://localhost:4000/ui/?page=logs), where to click, what fields to fill out, etc. along with the other commands to run in an ordered list, and I'll do it myself and post the screenshots after you make the PR
 
 If you ever write any human-facing text (pull requests, issues, commit messages, discussion posts, github comments, release notes, docs, etc.), always follow these guidelines to sound less AI-y:
 - don't use emojis

README.md

@@ -354,6 +354,8 @@ curl -X POST 'http://0.0.0.0:4000/v1/chat/completions' \
 | [Petals (`petals`)](https://docs.litellm.ai/docs/providers/petals) | ✅ | ✅ | ✅ |  |  |  |  |  |  |  |
 | [Pinstripes (`pinstripes`)](https://docs.litellm.ai/docs/providers/pinstripes) | ✅ | ✅ | ✅ |  |  |  |  |  |  |  |
 | [Predibase (`predibase`)](https://docs.litellm.ai/docs/providers/predibase) | ✅ | ✅ | ✅ |  |  |  |  |  |  |  |
+| [Qwen AI Platform (`qwen_ai_platform`)](https://docs.litellm.ai/docs/providers/qwencloud) | ✅ | ✅ | ✅ | ✅ | ✅ |  |  |  |  | ✅ |
+| [QwenCloud (`qwencloud`)](https://docs.litellm.ai/docs/providers/qwencloud) | ✅ | ✅ | ✅ | ✅ | ✅ |  |  |  |  | ✅ |
 | [Recraft (`recraft`)](https://docs.litellm.ai/docs/providers/recraft) |  |  |  |  | ✅ |  |  |  |  |  |
 | [Replicate (`replicate`)](https://docs.litellm.ai/docs/providers/replicate) | ✅ | ✅ | ✅ |  |  |  |  |  |  |  |
 | [Sagemaker Chat (`sagemaker_chat`)](https://docs.litellm.ai/docs/providers/aws_sagemaker) | ✅ | ✅ | ✅ |  |  |  |  |  |  |  |

enterprise/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "litellm-enterprise"
-version = "0.1.62"
+version = "0.1.63"
 description = "Package for LiteLLM Enterprise features"
 readme = "README.md"
 requires-python = ">=3.9"
@@ -26,7 +26,7 @@ required-version = ">=0.10.9"
 module-root = ""
 
 [tool.commitizen]
-version = "0.1.62"
+version = "0.1.63"
 version_files = [
     "pyproject.toml:^version",
     "../pyproject.toml:litellm-enterprise==",

litellm-proxy-extras/litellm_proxy_extras/utils.py

@@ -7,7 +7,8 @@ import subprocess
 import tempfile
 import time
 from pathlib import Path
-from typing import Optional
+from types import MappingProxyType
+from typing import Final, Optional
 
 from litellm_proxy_extras._logging import logger
 from litellm_proxy_extras.replica_identity import (
@@ -50,6 +51,17 @@ _SPEND_LOGS_PK_CLAUSE_RE = re.compile(
     re.IGNORECASE,
 )
 
+_PRISMA_ATTEMPTS: Final = 4
+
+_TRANSIENT_PRISMA_FAILURES: Final = MappingProxyType(
+    {
+        "deadlock detected": "a deadlock on the migration advisory lock (a concurrent migrate deploy)",
+        "P1001": "an unreachable database server",
+        "P1002": "a database server that timed out",
+    }
+)
+
+
 PARTITIONED_SPEND_LOGS_PUSH_ERROR = (
     "LiteLLM_SpendLogs is a partitioned table (see db_scripts/partition_spend_logs.sql), "
     "so its primary key must include the partition key (\"startTime\"). `prisma db push` "
@@ -274,6 +286,23 @@ class ProxyExtrasDBManager:
             env=prisma_env,
         )
 
+    @staticmethod
+    def _transient_prisma_failure(stderr: str) -> str | None:
+        """Why a failed prisma command is worth retrying, or None.
+
+        v1 retried every failure, so it absorbed a database that was not up yet
+        or another instance holding the migration lock. v2 fails fast, which is
+        right for a broken migration and wrong for these.
+        """
+        return next(
+            (
+                reason
+                for marker, reason in _TRANSIENT_PRISMA_FAILURES.items()
+                if marker in stderr
+            ),
+            None,
+        )
+
     @staticmethod
     def _is_permission_error(error_message: str) -> bool:
         """
@@ -655,7 +684,7 @@ class ProxyExtrasDBManager:
     @staticmethod
     def _setup_database_v2(use_migrate: bool) -> bool:
         """
-        v2 migration resolver (opt-in via --use_v2_migration_resolver).
+        v2 migration resolver (what the proxy CLI selects by default).
 
         Runs `prisma migrate deploy` and handles standard recovery paths
         (P3005 baseline, P3009/P3018 idempotent errors). Critically, it does
@@ -676,20 +705,46 @@ class ProxyExtrasDBManager:
             original_dir = os.getcwd()
             os.chdir(migrations_dir)
             try:
-                subprocess.run(
-                    [_get_prisma_command(), "db", "push", "--accept-data-loss"],
-                    timeout=prisma_command_timeout(),
-                    check=True,
-                    env=_get_prisma_env(),
+                for attempt in range(_PRISMA_ATTEMPTS):
+                    try:
+                        subprocess.run(
+                            [_get_prisma_command(), "db", "push", "--accept-data-loss"],
+                            timeout=prisma_command_timeout(),
+                            check=True,
+                            capture_output=True,
+                            text=True,
+                            env=_get_prisma_env(),
+                        )
+                        return True
+                    except subprocess.TimeoutExpired:
+                        logger.info(
+                            "prisma db push attempt %s timed out, retrying",
+                            attempt + 1,
+                        )
+                        time.sleep(random.randrange(5, 15))
+                    except subprocess.CalledProcessError as e:
+                        stderr = e.stderr or ""
+                        transient = ProxyExtrasDBManager._transient_prisma_failure(
+                            stderr
+                        )
+                        # Re-raise as RuntimeError so proxy_cli.py's
+                        # `except RuntimeError` catches it and exits cleanly.
+                        if transient is None or attempt == _PRISMA_ATTEMPTS - 1:
+                            raise RuntimeError(
+                                f"prisma db push failed.\n\nDetail: {e}"
+                                f"\n\nPrisma error:\n{stderr}"
+                            ) from e
+                        logger.info(
+                            "prisma db push attempt %s failed on %s, retrying. "
+                            "Prisma error:\n%s",
+                            attempt + 1,
+                            transient,
+                            stderr,
+                        )
+                        time.sleep(random.randrange(5, 15))
+                raise RuntimeError(
+                    f"prisma db push failed after {_PRISMA_ATTEMPTS} attempts."
                 )
-                return True
-            except (
-                subprocess.CalledProcessError,
-                subprocess.TimeoutExpired,
-            ) as e:
-                # Re-raise as RuntimeError so proxy_cli.py's
-                # `except RuntimeError` catches it and exits cleanly.
-                raise RuntimeError(f"prisma db push failed.\n\nDetail: {e}") from e
             finally:
                 os.chdir(original_dir)
 
@@ -699,7 +754,7 @@ class ProxyExtrasDBManager:
         original_dir = os.getcwd()
         os.chdir(migrations_dir)
         try:
-            for attempt in range(4):
+            for attempt in range(_PRISMA_ATTEMPTS):
                 try:
                     result = subprocess.run(
                         [_get_prisma_command(), "migrate", "deploy"],
@@ -814,16 +869,36 @@ class ProxyExtrasDBManager:
                             f"Manual intervention required.\n\nPrisma error:\n{stderr}"
                         ) from e
 
-                    raise RuntimeError(
-                        "Database migration failed and cannot be auto-recovered. "
-                        f"Manual intervention required.\n\nPrisma error:\n{stderr}"
-                    ) from e
+                    transient = ProxyExtrasDBManager._transient_prisma_failure(stderr)
+                    if transient is None:
+                        raise RuntimeError(
+                            "Database migration failed and cannot be auto-recovered. "
+                            f"Manual intervention required.\n\nPrisma error:\n{stderr}"
+                        ) from e
+
+                    if attempt == _PRISMA_ATTEMPTS - 1:
+                        raise RuntimeError(
+                            f"Database migration failed after "
+                            f"{_PRISMA_ATTEMPTS} attempts on {transient}. "
+                            "Check database connectivity and load."
+                            f"\n\nPrisma error:\n{stderr}"
+                        ) from e
+
+                    logger.info(
+                        "prisma migrate deploy attempt %s failed on %s, retrying. "
+                        "Prisma error:\n%s",
+                        attempt + 1,
+                        transient,
+                        stderr,
+                    )
+                    time.sleep(random.randrange(5, 15))
+                    continue
 
             raise RuntimeError(
-                "Database migration failed after 4 attempts (retry loop "
-                "exhausted by timeouts or repeated idempotent-recovery "
-                "continues). Check database connectivity, load, and "
-                "_prisma_migrations ledger state."
+                f"Database migration failed after {_PRISMA_ATTEMPTS} "
+                "attempts (retry loop exhausted by timeouts or repeated "
+                "idempotent-recovery continues). Check database connectivity, "
+                "load, and _prisma_migrations ledger state."
             )
         finally:
             os.chdir(original_dir)
@@ -871,10 +946,11 @@ class ProxyExtrasDBManager:
 
         Args:
             use_migrate: Whether to use prisma migrate instead of db push
-            use_v2_resolver: Opt into the v2 migration resolver (safer during
+            use_v2_resolver: Run the v2 migration resolver (safer during
                 rolling deploys; does not run the diff-and-force recovery
-                that causes schema thrashing). Defaults to False for
-                backwards compatibility.
+                that causes schema thrashing). Defaults to False here so
+                direct callers keep the old behavior; the proxy CLI passes
+                True, so the proxy's runtime default is v2.
 
         Returns:
             bool: True if setup was successful, False otherwise
@@ -892,7 +968,7 @@ class ProxyExtrasDBManager:
     @staticmethod
     def _run_migrations(use_migrate: bool, use_v2_resolver: bool) -> bool:
         if use_v2_resolver:
-            logger.info("Using v2 migration resolver (--use_v2_migration_resolver)")
+            logger.info("Using v2 migration resolver")
             return ProxyExtrasDBManager._setup_database_v2(use_migrate=use_migrate)
 
         schema_path = ProxyExtrasDBManager._get_prisma_dir() + "/schema.prisma"

litellm-proxy-extras/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "litellm-proxy-extras"
-version = "0.4.91"
+version = "0.4.92"
 description = "Additional files for the LiteLLM Proxy. Reduces the size of the main litellm package."
 readme = "README.md"
 requires-python = ">=3.9"
@@ -26,7 +26,7 @@ required-version = ">=0.10.9"
 module-root = ""
 
 [tool.commitizen]
-version = "0.4.91"
+version = "0.4.92"
 version_files = [
     "pyproject.toml:^version",
     "../pyproject.toml:litellm-proxy-extras==",

litellm-proxy-extras/tests/test_setup_database_fail_fast.py

@@ -1,242 +0,0 @@
-"""Regression tests for ProxyExtrasDBManager v2 migration resolver.
-
-The v2 resolver is opt-in via `--use_v2_migration_resolver` / the
-`use_v2_resolver=True` kwarg. These tests exercise the v2 path; the v1
-(default) behavior is unchanged from pre-fix.
-"""
-
-import subprocess
-from unittest.mock import patch
-
-import pytest
-
-from litellm_proxy_extras.utils import (
-    ProxyExtrasDBManager,
-    _max_migration_timestamp,
-    _migration_timestamp,
-)
-
-
-def _fake_migrate_deploy_failure(returncode: int, stderr: str):
-    def _run(*args, **kwargs):
-        raise subprocess.CalledProcessError(
-            returncode=returncode,
-            cmd=args[0],
-            stderr=stderr,
-            output="",
-        )
-
-    return _run
-
-
-def test_v2_p3018_permission_error_raises_runtime_error(monkeypatch, tmp_path):
-    """v2: a permission failure during migrate deploy raises RuntimeError."""
-    monkeypatch.setenv("DATABASE_URL", "postgresql://u:p@localhost:9/x")
-    monkeypatch.setattr(
-        ProxyExtrasDBManager, "_warn_if_db_ahead_of_head", lambda _: None
-    )
-    monkeypatch.setattr(ProxyExtrasDBManager, "_get_prisma_dir", lambda: str(tmp_path))
-    (tmp_path / "schema.prisma").write_text("// stub")
-
-    stderr = (
-        "Error: P3018\nMigration name: 20250326162113_baseline\n"
-        "Database error code: 42501\npermission denied for schema public"
-    )
-    with patch("subprocess.run", side_effect=_fake_migrate_deploy_failure(1, stderr)):
-        with pytest.raises(RuntimeError, match="permission"):
-            ProxyExtrasDBManager.setup_database(use_migrate=True, use_v2_resolver=True)
-
-
-def test_v2_non_idempotent_p3009_raises_runtime_error(monkeypatch, tmp_path):
-    """v2: a non-idempotent migration failure raises (no silent recovery)."""
-    monkeypatch.setenv("DATABASE_URL", "postgresql://u:p@localhost:9/x")
-    monkeypatch.setattr(
-        ProxyExtrasDBManager, "_warn_if_db_ahead_of_head", lambda _: None
-    )
-    monkeypatch.setattr(ProxyExtrasDBManager, "_get_prisma_dir", lambda: str(tmp_path))
-    (tmp_path / "schema.prisma").write_text("// stub")
-
-    stderr = (
-        "Error: P3009\nMigration `20260101000000_genuinely_broken` failed\n"
-        'Reason: syntax error at or near "BRKN" LINE 42'
-    )
-    with patch("subprocess.run", side_effect=_fake_migrate_deploy_failure(1, stderr)):
-        with pytest.raises(RuntimeError, match="cannot be auto-recovered"):
-            ProxyExtrasDBManager.setup_database(use_migrate=True, use_v2_resolver=True)
-
-
-def test_strip_prisma_query_params_removes_connection_limit():
-    """DATABASE_URLs with Prisma-specific params should be parseable by psycopg."""
-    url = "postgresql://u:p@h:5432/db?connection_limit=100&pool_timeout=60&sslmode=require"
-    stripped = ProxyExtrasDBManager._strip_prisma_query_params(url)
-    assert "connection_limit" not in stripped
-    assert "pool_timeout" not in stripped
-    assert "sslmode=require" in stripped
-
-
-def test_strip_prisma_query_params_passthrough_no_query():
-    """URLs without query strings are returned unchanged."""
-    url = "postgresql://u:p@h:5432/db"
-    assert ProxyExtrasDBManager._strip_prisma_query_params(url) == url
-
-
-def test_migration_timestamp_extracts_leading_digits():
-    assert _migration_timestamp("20260101000000_add_foo") == 20260101000000
-    assert _migration_timestamp("20250326162113_baseline") == 20250326162113
-
-
-def test_migration_timestamp_returns_zero_on_malformed():
-    assert _migration_timestamp("0_init") == 0
-    assert _migration_timestamp("not_a_migration") == 0
-
-
-def test_max_migration_timestamp():
-    names = {"20250326000000_a", "20260415000000_b", "20251115000000_c"}
-    assert _max_migration_timestamp(names) == 20260415000000
-
-
-def test_max_migration_timestamp_empty_set():
-    assert _max_migration_timestamp(set()) == 0
-
-
-def test_v1_default_still_calls_resolve_all_migrations(monkeypatch, tmp_path):
-    """v1 (default) continues to call _resolve_all_migrations on the happy path.
-
-    This is the existing buggy behavior — we're not fixing it in v1, only
-    offering v2 as opt-in. This test pins the default so that a future
-    inadvertent default flip is caught.
-    """
-    monkeypatch.setattr(ProxyExtrasDBManager, "_get_prisma_dir", lambda: str(tmp_path))
-    (tmp_path / "schema.prisma").write_text("// stub")
-
-    # Stub `prisma migrate deploy` to claim success with pending migrations
-    # applied, which is the code path that triggers the legacy post-migration
-    # sanity check (a call to _resolve_all_migrations).
-    class FakeResult:
-        stdout = "Applied migration.\n"
-        stderr = ""
-
-    def fake_run(cmd, *args, **kwargs):
-        return FakeResult()
-
-    resolve_called = {"n": 0}
-
-    def fake_resolve(*args, **kwargs):
-        resolve_called["n"] += 1
-
-    monkeypatch.setattr("subprocess.run", fake_run)
-    monkeypatch.setattr(ProxyExtrasDBManager, "_resolve_all_migrations", fake_resolve)
-
-    ok = ProxyExtrasDBManager.setup_database(use_migrate=True)  # v2 flag NOT set
-    assert ok is True
-    assert resolve_called["n"] == 1, "v1 default should still invoke the legacy path"
-
-
-def test_v2_db_push_wraps_subprocess_error_as_runtime_error(monkeypatch, tmp_path):
-    """v2: a failing `prisma db push` must raise RuntimeError, not leak
-    CalledProcessError past proxy_cli.py's `except RuntimeError`."""
-    monkeypatch.setattr(ProxyExtrasDBManager, "_get_prisma_dir", lambda: str(tmp_path))
-    (tmp_path / "schema.prisma").write_text("// stub")
-
-    stderr = "db push error"
-    with patch("subprocess.run", side_effect=_fake_migrate_deploy_failure(1, stderr)):
-        with pytest.raises(RuntimeError, match="prisma db push failed"):
-            ProxyExtrasDBManager.setup_database(use_migrate=False, use_v2_resolver=True)
-
-
-def test_v2_warn_ahead_of_head_swallows_db_errors(monkeypatch, tmp_path):
-    """_warn_if_db_ahead_of_head must never raise — it's informational.
-
-    Non-connection DB errors (e.g. InsufficientPrivilege from a user
-    without SELECT on _prisma_migrations) must be caught, not propagated.
-    """
-    import psycopg
-
-    monkeypatch.setenv("DATABASE_URL", "postgresql://u:p@localhost:9/x")
-    monkeypatch.setattr(ProxyExtrasDBManager, "_get_prisma_dir", lambda: str(tmp_path))
-    (tmp_path / "schema.prisma").write_text("// stub")
-
-    class _FakeConn:
-        def __enter__(self):
-            return self
-
-        def __exit__(self, *a):
-            return False
-
-        def execute(self, *a, **kw):
-            # Simulate an InsufficientPrivilege (subclass of DatabaseError).
-            raise psycopg.errors.InsufficientPrivilege("permission denied")
-
-    def _fake_connect(*a, **kw):
-        return _FakeConn()
-
-    monkeypatch.setattr("psycopg.connect", _fake_connect)
-
-    # Must not raise.
-    ProxyExtrasDBManager._warn_if_db_ahead_of_head(str(tmp_path))
-
-
-def test_v2_resolve_specific_migration_failure_raises_runtime_error(
-    monkeypatch, tmp_path
-):
-    """If marking a migration as applied fails inside P3009 idempotent
-    recovery, the subprocess error must be re-raised as RuntimeError so
-    proxy_cli.py catches it cleanly (instead of leaking CalledProcessError)."""
-    monkeypatch.setattr(
-        ProxyExtrasDBManager, "_warn_if_db_ahead_of_head", lambda _: None
-    )
-    monkeypatch.setattr(ProxyExtrasDBManager, "_get_prisma_dir", lambda: str(tmp_path))
-    (tmp_path / "schema.prisma").write_text("// stub")
-    monkeypatch.setattr(
-        ProxyExtrasDBManager, "_roll_back_migration", lambda *a, **kw: None
-    )
-
-    # First call: migrate deploy -> P3009 idempotent error.
-    # Recovery path tries _resolve_specific_migration; that also raises.
-    def _failing_resolve(*a, **kw):
-        raise subprocess.CalledProcessError(
-            returncode=1,
-            cmd="prisma migrate resolve --applied",
-            stderr="resolve failed",
-            output="",
-        )
-
-    monkeypatch.setattr(
-        ProxyExtrasDBManager, "_resolve_specific_migration", _failing_resolve
-    )
-
-    stderr = (
-        "Error: P3009\nMigration `20260101000000_some_migration` failed\n"
-        "relation already exists"
-    )
-    with patch("subprocess.run", side_effect=_fake_migrate_deploy_failure(1, stderr)):
-        with pytest.raises(
-            RuntimeError, match="Failed to mark migration .* as applied"
-        ):
-            ProxyExtrasDBManager.setup_database(use_migrate=True, use_v2_resolver=True)
-
-
-def test_v2_does_not_call_resolve_all_migrations(monkeypatch, tmp_path):
-    """v2 must never call _resolve_all_migrations — that's the bug it fixes."""
-    monkeypatch.setattr(
-        ProxyExtrasDBManager, "_warn_if_db_ahead_of_head", lambda _: None
-    )
-    monkeypatch.setattr(ProxyExtrasDBManager, "_get_prisma_dir", lambda: str(tmp_path))
-    (tmp_path / "schema.prisma").write_text("// stub")
-
-    class FakeResult:
-        stdout = "Applied migration.\n"
-        stderr = ""
-
-    monkeypatch.setattr("subprocess.run", lambda *a, **kw: FakeResult())
-
-    resolve_called = {"n": 0}
-    monkeypatch.setattr(
-        ProxyExtrasDBManager,
-        "_resolve_all_migrations",
-        lambda *a, **kw: resolve_called.__setitem__("n", resolve_called["n"] + 1),
-    )
-
-    ok = ProxyExtrasDBManager.setup_database(use_migrate=True, use_v2_resolver=True)
-    assert ok is True
-    assert resolve_called["n"] == 0, "v2 must not invoke the diff-and-force recovery"

litellm/__init__.py

@@ -659,6 +659,8 @@ aiml_models: Set = set()
 deepgram_models: Set = set()
 elevenlabs_models: Set = set()
 dashscope_models: Set = set()
+qwencloud_models: Set = set()
+qwen_ai_platform_models: Set = set()
 moonshot_models: Set = set()
 publicai_models: Set = set()
 darkbloom_models: Set = set()
@@ -909,6 +911,10 @@ def _populate_provider_model_sets(model_cost_map: Dict) -> None:
             heroku_models.add(key)
         elif value.get("litellm_provider") == "dashscope":
             dashscope_models.add(key)
+        elif value.get("litellm_provider") == "qwencloud":
+            qwencloud_models.add(key)
+        elif value.get("litellm_provider") == "qwen_ai_platform":
+            qwen_ai_platform_models.add(key)
         elif value.get("litellm_provider") == "modelscope":
             modelscope_models.add(key)
         elif value.get("litellm_provider") == "moonshot":
@@ -1072,6 +1078,8 @@ model_list = list(
     | deepgram_models
     | elevenlabs_models
     | dashscope_models
+    | qwencloud_models
+    | qwen_ai_platform_models
     | moonshot_models
     | publicai_models
     | darkbloom_models
@@ -1178,6 +1186,8 @@ def _build_models_by_provider() -> dict:
         "elevenlabs": elevenlabs_models,
         "heroku": heroku_models,
         "dashscope": dashscope_models,
+        "qwencloud": qwencloud_models,
+        "qwen_ai_platform": qwen_ai_platform_models,
         "modelscope": modelscope_models,
         "moonshot": moonshot_models,
         "publicai": publicai_models,
@@ -2014,6 +2024,24 @@ if TYPE_CHECKING:
     from .llms.dashscope.rerank.transformation import (
         DashScopeRerankConfig as DashScopeRerankConfig,
     )
+    from .llms.dashscope.qwencloud import (
+        QwenCloudChatConfig as QwenCloudChatConfig,
+    )
+    from .llms.dashscope.qwencloud import (
+        QwenCloudEmbeddingConfig as QwenCloudEmbeddingConfig,
+    )
+    from .llms.dashscope.qwencloud import (
+        QwenCloudRerankConfig as QwenCloudRerankConfig,
+    )
+    from .llms.dashscope.qwen_ai_platform import (
+        QwenAIPlatformChatConfig as QwenAIPlatformChatConfig,
+    )
+    from .llms.dashscope.qwen_ai_platform import (
+        QwenAIPlatformEmbeddingConfig as QwenAIPlatformEmbeddingConfig,
+    )
+    from .llms.dashscope.qwen_ai_platform import (
+        QwenAIPlatformRerankConfig as QwenAIPlatformRerankConfig,
+    )
     from .llms.modelscope.chat.transformation import (
         ModelScopeChatConfig as ModelScopeChatConfig,
     )

litellm/_lazy_imports_registry.py

@@ -310,6 +310,8 @@ LLM_CONFIG_NAMES: Final = (
     "GigaChatConfig",
     "GigaChatEmbeddingConfig",
     "DashScopeChatConfig",
+    "QwenCloudChatConfig",
+    "QwenAIPlatformChatConfig",
     "ModelScopeChatConfig",
     "MoonshotChatConfig",
     "DockerModelRunnerChatConfig",
@@ -1172,6 +1174,14 @@ _LLM_CONFIGS_IMPORT_MAP: Final = {
         ".llms.dashscope.chat.transformation",
         "DashScopeChatConfig",
     ),
+    "QwenCloudChatConfig": (
+        ".llms.dashscope.qwencloud",
+        "QwenCloudChatConfig",
+    ),
+    "QwenAIPlatformChatConfig": (
+        ".llms.dashscope.qwen_ai_platform",
+        "QwenAIPlatformChatConfig",
+    ),
     "GDCGeminiConfig": (
         ".llms.gdc.chat.transformation",
         "GDCGeminiConfig",

litellm/constants.py

@@ -630,6 +630,8 @@ LITELLM_CHAT_PROVIDERS: Final = [
     "nscale",
     "nebius",
     "dashscope",
+    "qwencloud",
+    "qwen_ai_platform",
     "modelscope",
     "moonshot",
     "publicai",
@@ -799,6 +801,7 @@ openai_compatible_endpoints: Final[list] = [
     "inference.api.nscale.com/v1",
     "api.studio.nebius.ai/v1",
     "https://dashscope-intl.aliyuncs.com/compatible-mode/v1",
+    "https://dashscope.aliyuncs.com/compatible-mode/v1",
     "https://api-inference.modelscope.cn/v1",
     "https://api.moonshot.ai/v1",
     "https://api.publicai.co/v1",
@@ -872,6 +875,8 @@ openai_compatible_providers: Final[list] = [
     "nscale",
     "nebius",
     "dashscope",
+    "qwencloud",
+    "qwen_ai_platform",
     "modelscope",
     "moonshot",
     "v0",
@@ -902,6 +907,8 @@ openai_text_completion_compatible_providers: Final[list] = [  # providers that s
     "featherless_ai",
     "nebius",
     "dashscope",
+    "qwencloud",
+    "qwen_ai_platform",
     "modelscope",
     "moonshot",
     "publicai",
@@ -1109,7 +1116,7 @@ nebius_models: Final[set] = set(
     ]
 )
 
-dashscope_models: Final[set] = set(
+dashscope_models: Final[frozenset] = frozenset(
     [
         "qwen-turbo",
         "qwen-plus",
@@ -1124,6 +1131,10 @@ dashscope_models: Final[set] = set(
     ]
 )
 
+qwencloud_models: Final[frozenset] = frozenset(dashscope_models)
+
+qwen_ai_platform_models: Final[frozenset] = frozenset(dashscope_models)
+
 nebius_embedding_models: Final[set] = set(
     [
         "BAAI/bge-en-icl",

litellm/cost_calculator.py

@@ -641,12 +641,12 @@ def cost_per_token(
         return xai_cost_per_token(model=model, usage=usage_block)
     elif custom_llm_provider == "lemonade":
         return lemonade_cost_per_token(model=model, usage=usage_block)
-    elif custom_llm_provider == "dashscope":
+    elif custom_llm_provider in ("dashscope", "qwencloud", "qwen_ai_platform"):
         from litellm.llms.dashscope.cost_calculator import (
             cost_per_token as dashscope_cost_per_token,
         )
 
-        return dashscope_cost_per_token(model=model, usage=usage_block)
+        return dashscope_cost_per_token(model=model, usage=usage_block, custom_llm_provider=custom_llm_provider)
     elif custom_llm_provider == "azure_ai":
         return azure_ai_cost_per_token(
             model=model,

litellm/images/main.py

@@ -386,6 +386,8 @@ def image_generation(
             litellm.LlmProviders.VERTEX_AI,
             litellm.LlmProviders.OPENROUTER,
             litellm.LlmProviders.DASHSCOPE,
+            litellm.LlmProviders.QWENCLOUD,
+            litellm.LlmProviders.QWEN_AI_PLATFORM,
         ):
             if image_generation_config is None:
                 raise ValueError(f"image generation config is not supported for {custom_llm_provider}")

litellm/integrations/websearch_interception/handler.py

@@ -416,25 +416,15 @@ class WebSearchInterceptionLogger(CustomLogger):
         if not tools:
             return None
 
-        is_responses_call: Final = call_type in (CallTypes.responses, CallTypes.aresponses)
-        has_websearch: Final = (
-            any(is_web_search_tool_responses(tool) for tool in tools)
-            if is_responses_call
-            else any(is_web_search_tool(tool) for tool in tools)
-        )
+        if call_type in (CallTypes.responses, CallTypes.aresponses):
+            return self._convert_responses_tools(kwargs=kwargs, tools=tools)
+
+        # Check if any tool is a web search tool (native or already LiteLLM standard)
+        has_websearch: Final = any(is_web_search_tool(t) for t in tools)
+
         if not has_websearch:
             return None
 
-        if self.search_tool_name:
-            try:
-                from litellm.proxy.proxy_server import llm_router
-            except ImportError:
-                llm_router = None
-            self._select_search_tool_from_router(llm_router=llm_router)
-
-        if is_responses_call:
-            return self._convert_responses_tools(kwargs=kwargs, tools=tools)
-
         verbose_logger.debug("WebSearchInterception: Converting native web_search tools to LiteLLM standard")
 
         # If the client sent an Anthropic-native web_search_* tool, mark the
@@ -1641,36 +1631,34 @@ class WebSearchInterceptionLogger(CustomLogger):
         return None
 
     def _select_search_tool_from_router(self, llm_router: object) -> "_SearchToolConfig | None":
-        search_tools: Final = list(getattr(llm_router, "search_tools", []) or [])
+        if llm_router is None or not hasattr(llm_router, "search_tools"):
+            return None
+        search_tools: Final = tuple(getattr(llm_router, "search_tools", None) or ())
         return self._select_search_tool_from_list(search_tools=search_tools, source="router")
 
     def _select_search_tool_from_list(
         self,
-        search_tools: list[_SearchToolConfig],
+        search_tools: Sequence[_SearchToolConfig],
         source: str,
     ) -> "_SearchToolConfig | None":
         if self.search_tool_name:
-            matching_tools = [tool for tool in search_tools if tool.get("search_tool_name") == self.search_tool_name]
-            if not matching_tools:
-                raise ValueError(f"Configured search tool '{self.search_tool_name}' was not found")
-
-            selected_tool: Final = matching_tools[0]
-            litellm_params: Final = selected_tool.get("litellm_params")
-            selected_search_provider: Final = (
-                litellm_params.get("search_provider") if isinstance(litellm_params, Mapping) else None
+            matching_tools: Final = tuple(
+                tool for tool in search_tools if tool.get("search_tool_name") == self.search_tool_name
             )
-            if not isinstance(selected_search_provider, str) or not selected_search_provider.strip():
-                raise ValueError(
-                    f"Configured search tool '{self.search_tool_name}' does not define a valid search provider"
+            if matching_tools:
+                search_provider = (matching_tools[0].get("litellm_params", {}) or {}).get("search_provider")
+                verbose_logger.debug(
+                    "WebSearchInterception: Found search tool '%s' from %s with provider '%s'",
+                    self.search_tool_name,
+                    source,
+                    search_provider,
                 )
-
+                return matching_tools[0]
             verbose_logger.debug(
-                "WebSearchInterception: Found search tool '%s' from %s with provider '%s'",
+                "WebSearchInterception: Search tool '%s' not found in %s, falling back to first available or perplexity",
                 self.search_tool_name,
                 source,
-                selected_search_provider,
             )
-            return selected_tool
 
         if search_tools:
             first_tool: Final = search_tools[0]

litellm/litellm_core_utils/get_llm_provider_logic.py

@@ -536,6 +536,14 @@ def get_llm_provider(
             )
 
 
+def _dashscope_family_chat_config(custom_llm_provider: str) -> "litellm.DashScopeChatConfig":
+    if custom_llm_provider == "qwencloud":
+        return litellm.QwenCloudChatConfig()
+    if custom_llm_provider == "qwen_ai_platform":
+        return litellm.QwenAIPlatformChatConfig()
+    return litellm.DashScopeChatConfig()
+
+
 def _get_openai_compatible_provider_info(
     model: str,
     api_base: str | None,
@@ -785,11 +793,11 @@ def _get_openai_compatible_provider_info(
             api_base,
             dynamic_api_key,
         ) = litellm.HerokuChatConfig()._get_openai_compatible_provider_info(api_base, api_key)
-    elif custom_llm_provider == "dashscope":
+    elif custom_llm_provider in ("dashscope", "qwencloud", "qwen_ai_platform"):
         (
             api_base,
             dynamic_api_key,
-        ) = litellm.DashScopeChatConfig()._get_openai_compatible_provider_info(api_base, api_key)
+        ) = _dashscope_family_chat_config(custom_llm_provider)._get_openai_compatible_provider_info(api_base, api_key)
     elif custom_llm_provider == "modelscope":
         (
             api_base,

litellm/litellm_core_utils/streaming_handler.py

@@ -862,6 +862,8 @@ class CustomStreamWrapper:
         model_response: Final = ModelResponseStream(**args)
         if self.response_id is not None:
             model_response.id = self.response_id
+        elif model_response.id:
+            self.response_id = model_response.id
         if self.system_fingerprint is not None:
             model_response.system_fingerprint = self.system_fingerprint
 

litellm/llms/anthropic/chat/transformation.py

@@ -1994,19 +1994,35 @@ class AnthropicConfig(AnthropicModelInfo, BaseConfig):
         return data
 
     def _apply_output_config(self, data: dict, model: str, optional_params: dict) -> None:
-        """Validate and apply output_config to the request data."""
+        """Validate and apply output_config to the request data.
+
+        The ``drop_params`` gate here is an effort gate: ``format`` is a
+        structured-output field, not an effort field, so it survives the drop
+        and is vetted where it is consumed (the map's
+        ``supports_native_structured_output`` flag on emission paths).
+        """
         if "output_config" not in optional_params:
             return
         output_config: Final = optional_params.get("output_config")
         if not output_config or not isinstance(output_config, dict):
             return
-        if litellm.drop_params is True and not self._model_supports_effort_param(model, self._resolved_provider):
+        if (
+            litellm.drop_params is True
+            and any(key != "format" for key in output_config)
+            and not self._model_supports_effort_param(model, self._resolved_provider)
+        ):
             litellm.verbose_logger.warning(
                 DROP_UNSUPPORTED_OUTPUT_CONFIG_WARNING,
                 model,
             )
-            optional_params.pop("output_config", None)
-            data.pop("output_config", None)
+            preserved_format: Final = output_config.get("format")
+            if preserved_format is None:
+                optional_params.pop("output_config", None)
+                data.pop("output_config", None)
+                return
+            format_only: Final = {"format": preserved_format}  # mutable-ok: json body
+            optional_params["output_config"] = format_only  # rebind-ok: out-param store
+            data["output_config"] = format_only  # rebind-ok: out-param store
             return
         effort: Final = output_config.get("effort")
         valid_efforts: Final = ["high", "medium", "low", "xhigh", "max"]

litellm/llms/bedrock/chat/invoke_transformations/anthropic_claude3_transformation.py

@@ -3,7 +3,6 @@ from typing import TYPE_CHECKING, Any, Final
 import httpx
 
 from litellm.anthropic_beta_headers_manager import filter_and_transform_beta_headers
-from litellm.litellm_core_utils.litellm_logging import verbose_logger
 from litellm.litellm_core_utils.prompt_templates.factory import (
     convert_to_anthropic_image_obj,
 )
@@ -16,17 +15,16 @@ from litellm.llms.bedrock.chat.invoke_transformations.base_invoke_transformation
     AmazonInvokeConfig,
 )
 from litellm.llms.bedrock.common_utils import (
-    convert_bedrock_invoke_output_format_to_inline_schema,
+    apply_bedrock_invoke_structured_output,
     get_anthropic_beta_from_headers,
     normalize_bedrock_opus_output_config_effort,
     normalize_custom_field_on_tools,
     normalize_tool_input_schema_types_for_bedrock_invoke,
-    pop_bedrock_invoke_output_config_format,
+    strip_unsupported_bedrock_invoke_output_config_keys,
 )
 from litellm.types.llms.anthropic import ANTHROPIC_TOOL_SEARCH_BETA_HEADER
 from litellm.types.llms.openai import AllMessageValues
 from litellm.types.utils import ModelResponse
-from litellm.utils import _supports_factory
 
 if TYPE_CHECKING:
     import tiktoken
@@ -212,36 +210,14 @@ class AmazonAnthropicClaudeConfig(AmazonInvokeConfig, AnthropicConfig):
         anthropic_request.pop("model", None)
         anthropic_request.pop("stream", None)
         anthropic_request.pop("stream_chunk_size", None)
-        output_format: Final = anthropic_request.pop("output_format", None)
-        output_config_format: Final = pop_bedrock_invoke_output_config_format(anthropic_request)
-        if output_format:
-            convert_bedrock_invoke_output_format_to_inline_schema(
-                output_format=output_format,
-                request_body=anthropic_request,
-            )
-        elif output_config_format:
-            convert_bedrock_invoke_output_format_to_inline_schema(
-                output_format=output_config_format,
-                request_body=anthropic_request,
-            )
-        if not (
-            _supports_factory(
-                model=model,
-                custom_llm_provider="bedrock",
-                key="supports_output_config",
-            )
-            or AnthropicConfig._model_supports_effort_param(model, "bedrock")
-        ):
-            if anthropic_request.pop("output_config", None) is not None:
-                verbose_logger.warning(
-                    "Bedrock Invoke: stripping unsupported `output_config` for "
-                    "model=%s — neither `supports_output_config` nor any "
-                    "`supports_*_reasoning_effort` flag is set in "
-                    "model_prices_and_context_window.json. Add the capability "
-                    "flag to the model JSON entry if this model accepts "
-                    "`output_config`.",
-                    model,
-                )
+        apply_bedrock_invoke_structured_output(
+            model=model,
+            request_body=anthropic_request,
+        )
+        strip_unsupported_bedrock_invoke_output_config_keys(
+            model=model,
+            request_body=anthropic_request,
+        )
         if "anthropic_version" not in anthropic_request:
             anthropic_request["anthropic_version"] = self.anthropic_version
 

litellm/llms/bedrock/common_utils.py

@@ -177,6 +177,95 @@ def convert_bedrock_invoke_output_format_to_inline_schema(
     request_body["messages"] = new_messages
 
 
+def _bedrock_model_supports(model: str, key: str) -> bool:
+    from litellm.utils import _supports_factory
+
+    return _supports_factory(model=model, custom_llm_provider="bedrock", key=key)
+
+
+def apply_bedrock_invoke_structured_output(
+    model: str,
+    request_body: dict[str, object],  # mutable-ok: edited in place like siblings
+) -> None:
+    """
+    Route Anthropic structured-output params to what the Bedrock model supports.
+
+    Consumes the legacy top-level ``output_format`` and the newer
+    ``output_config.format``, keeping the pre-existing precedence of the legacy
+    field when a request carries both. Models flagged
+    ``supports_native_structured_output`` in the model map get the schema
+    forwarded as ``output_config.format``, which Bedrock relays to the model for
+    enforced structured output. For every other model the schema is inlined into
+    the last user message as best-effort text, with a warning because nothing
+    enforces it.
+    """
+    legacy_output_format: Final = request_body.pop("output_format", None)
+    output_config_format: Final = pop_bedrock_invoke_output_config_format(request_body)
+    schema_format: Final = legacy_output_format if isinstance(legacy_output_format, dict) else output_config_format
+    if schema_format is None:
+        return
+
+    if _bedrock_model_supports(model, "supports_native_structured_output"):
+        existing_output_config: Final = request_body.get("output_config")
+        if isinstance(existing_output_config, dict):
+            existing_output_config["format"] = schema_format
+        else:
+            request_body["output_config"] = {"format": schema_format}  # rebind-ok: out-param  # mutable-ok: json
+        return
+
+    verbose_logger.warning(
+        "Bedrock Invoke: model=%s does not advertise `supports_native_structured_output` "
+        "in model_prices_and_context_window.json, so the JSON schema was inlined into "
+        "the last user message and is NOT enforced by the model.",
+        model,
+    )
+    convert_bedrock_invoke_output_format_to_inline_schema(
+        output_format=schema_format,
+        request_body=request_body,
+    )
+
+
+def strip_unsupported_bedrock_invoke_output_config_keys(
+    model: str,
+    request_body: dict[str, object],  # mutable-ok: edited in place like siblings
+) -> None:
+    """
+    Drop ``output_config`` keys the Bedrock model does not accept.
+
+    ``format`` survives unconditionally: it is only attached for models whose map
+    entry advertises ``supports_native_structured_output``. Effort-bearing keys
+    survive only when the map flags ``supports_output_config`` or a
+    ``supports_*_reasoning_effort`` tier; otherwise they are dropped with a
+    warning so Bedrock does not reject the request.
+    """
+    from litellm.llms.anthropic.chat.transformation import AnthropicConfig
+
+    output_config: Final = request_body.get("output_config")
+    if not isinstance(output_config, dict):
+        return
+    if all(key == "format" for key in output_config):
+        return
+    if _bedrock_model_supports(model, "supports_output_config") or AnthropicConfig._model_supports_effort_param(
+        model, "bedrock"
+    ):
+        return
+
+    verbose_logger.warning(
+        "Bedrock Invoke: stripping unsupported `output_config` keys for "
+        "model=%s: neither `supports_output_config` nor any "
+        "`supports_*_reasoning_effort` flag is set in "
+        "model_prices_and_context_window.json. Add the capability "
+        "flag to the model JSON entry if this model accepts "
+        "`output_config`.",
+        model,
+    )
+    preserved_format: Final = output_config.get("format")
+    if preserved_format is None:
+        request_body.pop("output_config", None)
+    else:
+        request_body["output_config"] = {"format": preserved_format}  # rebind-ok: out-param  # mutable-ok: json
+
+
 def normalize_custom_field_on_tools(request_body: dict) -> None:
     """
     Drop the ``custom`` field from each tool, first hoisting a boolean

litellm/llms/bedrock/messages/invoke_transformations/anthropic_claude3_transformation.py

@@ -29,14 +29,14 @@ from litellm.llms.bedrock.chat.invoke_transformations.base_invoke_transformation
     AmazonInvokeConfig,
 )
 from litellm.llms.bedrock.common_utils import (
-    convert_bedrock_invoke_output_format_to_inline_schema,
+    apply_bedrock_invoke_structured_output,
     ensure_bedrock_anthropic_messages_tool_names,
     get_anthropic_beta_from_headers,
     is_claude_4_5_on_bedrock,
     normalize_bedrock_opus_output_config_effort,
     normalize_custom_field_on_tools,
     normalize_tool_input_schema_types_for_bedrock_invoke,
-    pop_bedrock_invoke_output_config_format,
+    strip_unsupported_bedrock_invoke_output_config_keys,
 )
 from litellm.llms.bedrock.request_metadata import (
     bedrock_request_metadata_headers,
@@ -51,7 +51,6 @@ from litellm.types.llms.openai import AllMessageValues
 from litellm.types.router import GenericLiteLLMParams
 from litellm.types.utils import GenericStreamingChunk, ModelResponseStream
 from litellm.types.utils import GenericStreamingChunk as GChunk
-from litellm.utils import _supports_factory
 
 if TYPE_CHECKING:
     from litellm.litellm_core_utils.litellm_logging import Logging as _LiteLLMLoggingObj
@@ -708,52 +707,25 @@ class AmazonAnthropicClaudeMessagesConfig(
         # 4. Remove `ttl` field from cache_control in messages (Bedrock doesn't support it for older models)
         self._remove_ttl_from_cache_control(anthropic_messages_request=anthropic_messages_request, model=model)
 
-        # 5. Convert structured-output params to inline schema.
-        # Bedrock Invoke doesn't support top-level `output_format`; its
-        # accepted `output_config` subset is also narrower than Anthropic's, so
-        # consume the newer `output_config.format` shape here instead of
-        # forwarding it as an unknown nested key.
+        # 5. Route structured-output params (`output_format` /
+        # `output_config.format`) to native enforcement or the inline-schema
+        # fallback, then strip `output_config` keys the model does not accept.
+        # Ref: https://github.com/BerriAI/litellm/issues/22797
         existing_output_config: Final = anthropic_messages_request.get("output_config")
         if isinstance(existing_output_config, dict):
             anthropic_messages_request["output_config"] = dict(existing_output_config)
-        output_format: Final = anthropic_messages_request.pop("output_format", None)
-        output_config_format: Final = pop_bedrock_invoke_output_config_format(anthropic_messages_request)
-        if output_format:
-            convert_bedrock_invoke_output_format_to_inline_schema(
-                output_format=output_format,
-                request_body=anthropic_messages_request,
-            )
-        elif output_config_format:
-            convert_bedrock_invoke_output_format_to_inline_schema(
-                output_format=output_config_format,
-                request_body=anthropic_messages_request,
-            )
+        apply_bedrock_invoke_structured_output(
+            model=model,
+            request_body=anthropic_messages_request,
+        )
         normalize_bedrock_opus_output_config_effort(
             model=model,
             output_config=anthropic_messages_request.get("output_config"),
         )
-
-        # 5a. Bedrock Invoke supports output_config (effort) for Claude 4.6+ models,
-        # but older models do not — strip it to avoid request rejection.
-        # Ref: https://github.com/BerriAI/litellm/issues/22797
-        if not (
-            _supports_factory(
-                model=model,
-                custom_llm_provider="bedrock",
-                key="supports_output_config",
-            )
-            or AnthropicConfig._model_supports_effort_param(model, "bedrock")
-        ):
-            if anthropic_messages_request.pop("output_config", None) is not None:
-                verbose_logger.warning(
-                    "Bedrock Invoke: stripping unsupported `output_config` for "
-                    "model=%s — neither `supports_output_config` nor any "
-                    "`supports_*_reasoning_effort` flag is set in "
-                    "model_prices_and_context_window.json. Add the capability "
-                    "flag to the model JSON entry if this model accepts "
-                    "`output_config`.",
-                    model,
-                )
+        strip_unsupported_bedrock_invoke_output_config_keys(
+            model=model,
+            request_body=anthropic_messages_request,
+        )
 
         # 5b. Hoist `custom.defer_loading` then drop `custom` (Bedrock doesn't support it)
         # Ref: https://github.com/BerriAI/litellm/issues/22847
@@ -774,9 +746,11 @@ class AmazonAnthropicClaudeMessagesConfig(
         if filtered_betas:
             anthropic_messages_request["anthropic_beta"] = filtered_betas
 
+        remaining_output_config: Final = anthropic_messages_request.get("output_config")
         if (
             litellm.drop_params is True
-            and "output_config" in anthropic_messages_request
+            and isinstance(remaining_output_config, dict)
+            and any(key != "format" for key in remaining_output_config)
             and not AnthropicConfig._model_supports_effort_param(model, "bedrock")
         ):
             verbose_logger.warning(

litellm/llms/dashscope/chat/transformation.py

@@ -54,6 +54,9 @@ class DashScopeChatConfig(OpenAIGPTConfig):
         dynamic_api_key: Final = api_key or get_secret_str("DASHSCOPE_API_KEY")
         return api_base, dynamic_api_key
 
+    def _resolve_chat_api_base(self, api_base: str | None) -> str:
+        return api_base or "https://dashscope.aliyuncs.com/compatible-mode/v1"
+
     def get_complete_url(
         self,
         api_base: str | None,
@@ -66,10 +69,7 @@ class DashScopeChatConfig(OpenAIGPTConfig):
         """
         If api_base is not provided, use the default DashScope /chat/completions endpoint.
         """
-        if not api_base:
-            api_base = "https://dashscope.aliyuncs.com/compatible-mode/v1"
-
-        if not api_base.endswith("/chat/completions"):
-            api_base = f"{api_base}/chat/completions"
-
-        return api_base
+        resolved_api_base: Final = self._resolve_chat_api_base(api_base)
+        if resolved_api_base.endswith("/chat/completions"):
+            return resolved_api_base
+        return f"{resolved_api_base}/chat/completions"

litellm/llms/dashscope/common_utils.py

@@ -2,9 +2,89 @@
 Common utilities for the DashScope LLM provider.
 """
 
+from typing import TYPE_CHECKING
+
 import httpx
 
 from litellm.llms.base_llm.chat.transformation import BaseLLMException
+from litellm.secret_managers.main import get_secret_str
+
+if TYPE_CHECKING:
+    from litellm.llms.base_llm.embedding.transformation import BaseEmbeddingConfig
+    from litellm.llms.base_llm.image_generation.transformation import (
+        BaseImageGenerationConfig,
+    )
+    from litellm.llms.base_llm.rerank.transformation import BaseRerankConfig
+
+
+def get_dashscope_family_embedding_config(custom_llm_provider: str) -> "BaseEmbeddingConfig":
+    if custom_llm_provider == "qwencloud":
+        from litellm.llms.dashscope.qwencloud import QwenCloudEmbeddingConfig
+
+        return QwenCloudEmbeddingConfig()
+    if custom_llm_provider == "qwen_ai_platform":
+        from litellm.llms.dashscope.qwen_ai_platform import (
+            QwenAIPlatformEmbeddingConfig,
+        )
+
+        return QwenAIPlatformEmbeddingConfig()
+    from litellm.llms.dashscope.embed.transformation import DashScopeEmbeddingConfig
+
+    return DashScopeEmbeddingConfig()
+
+
+def get_dashscope_family_rerank_config(custom_llm_provider: str) -> "BaseRerankConfig":
+    if custom_llm_provider == "qwencloud":
+        from litellm.llms.dashscope.qwencloud import QwenCloudRerankConfig
+
+        return QwenCloudRerankConfig()
+    if custom_llm_provider == "qwen_ai_platform":
+        from litellm.llms.dashscope.qwen_ai_platform import QwenAIPlatformRerankConfig
+
+        return QwenAIPlatformRerankConfig()
+    from litellm.llms.dashscope.rerank.transformation import DashScopeRerankConfig
+
+    return DashScopeRerankConfig()
+
+
+def get_dashscope_family_image_generation_config(
+    custom_llm_provider: str,
+) -> "BaseImageGenerationConfig":
+    if custom_llm_provider == "qwencloud":
+        from litellm.llms.dashscope.qwencloud import QwenCloudImageGenerationConfig
+
+        return QwenCloudImageGenerationConfig()
+    if custom_llm_provider == "qwen_ai_platform":
+        from litellm.llms.dashscope.qwen_ai_platform import (
+            QwenAIPlatformImageGenerationConfig,
+        )
+
+        return QwenAIPlatformImageGenerationConfig()
+    from litellm.llms.dashscope.image_generation.transformation import (
+        DashScopeImageGenerationConfig,
+    )
+
+    return DashScopeImageGenerationConfig()
+
+
+def resolve_dashscope_family_api_key(custom_llm_provider: str, api_key: str | None) -> str | None:
+    if custom_llm_provider == "dashscope":
+        return api_key or get_secret_str("DASHSCOPE_API_KEY")
+    return api_key or get_secret_str(f"{custom_llm_provider.upper()}_API_KEY") or get_secret_str("DASHSCOPE_API_KEY")
+
+
+def missing_dashscope_family_key_message(custom_llm_provider: str) -> str:
+    if custom_llm_provider == "qwencloud":
+        return (
+            "Missing API key for QwenCloud. Set QWENCLOUD_API_KEY or "
+            "DASHSCOPE_API_KEY environment variable or pass api_key parameter."
+        )
+    if custom_llm_provider == "qwen_ai_platform":
+        return (
+            "Missing API key for Qwen AI Platform. Set QWEN_AI_PLATFORM_API_KEY or "
+            "DASHSCOPE_API_KEY environment variable or pass api_key parameter."
+        )
+    return "Missing API key for DashScope. Set DASHSCOPE_API_KEY environment variable or pass api_key parameter."
 
 
 class DashScopeError(BaseLLMException):

litellm/llms/dashscope/cost_calculator.py

@@ -110,7 +110,7 @@ def _calculate_completion_cost(
     return (breakdown.completion_tokens * output_cost) + (breakdown.reasoning_tokens * reasoning_cost)
 
 
-def cost_per_token(model: str, usage: Usage) -> tuple[float, float]:
+def cost_per_token(model: str, usage: Usage, custom_llm_provider: str = "dashscope") -> tuple[float, float]:
     """
     Calculate cost per token for Dashscope models.
 
@@ -119,11 +119,12 @@ def cost_per_token(model: str, usage: Usage) -> tuple[float, float]:
     Args:
         model: Model name without provider prefix
         usage: LiteLLM Usage block
+        custom_llm_provider: The provider id the request resolved to; dashscope or one of its brand aliases
 
     Returns:
         Tuple[float, float] - (prompt_cost_in_usd, completion_cost_in_usd)
     """
-    model_info: Final = get_model_info(model=model, custom_llm_provider="dashscope")
+    model_info: Final = get_model_info(model=model, custom_llm_provider=custom_llm_provider)
     breakdown: Final = _extract_token_breakdown(usage)
     raw_tiers: Final = model_info.get("tiered_pricing")
     tiered_pricing: Final = raw_tiers if isinstance(raw_tiers, list) else None

litellm/llms/dashscope/embed/transformation.py

@@ -62,6 +62,17 @@ class DashScopeEmbeddingConfig(BaseEmbeddingConfig):
             # for drop_params=False before this method is called.
         return optional_params
 
+    def _resolve_api_key(self, api_key: str | None) -> str:
+        resolved_api_key: Final = api_key if api_key is not None else get_secret_str("DASHSCOPE_API_KEY")
+        if resolved_api_key is None:
+            raise ValueError(
+                "DashScope API key is required. Set 'DASHSCOPE_API_KEY' env var or pass api_key explicitly."
+            )
+        return resolved_api_key
+
+    def _resolve_embedding_api_base(self, api_base: str | None) -> str:
+        return api_base or get_secret_str("DASHSCOPE_API_BASE") or DEFAULT_API_BASE
+
     def validate_environment(
         self,
         headers: dict,
@@ -72,17 +83,11 @@ class DashScopeEmbeddingConfig(BaseEmbeddingConfig):
         api_key: str | None = None,
         api_base: str | None = None,
     ) -> dict:
-        if api_key is None:
-            api_key = get_secret_str("DASHSCOPE_API_KEY")
-        if api_key is None:
-            raise ValueError(
-                "DashScope API key is required. Set 'DASHSCOPE_API_KEY' env var or pass api_key explicitly."
-            )
-        default_headers: Final = {
+        return {
             "Content-Type": "application/json",
-            "Authorization": f"Bearer {api_key}",
+            "Authorization": f"Bearer {self._resolve_api_key(api_key)}",
+            **headers,
         }
-        return {**default_headers, **headers}
 
     def get_complete_url(
         self,
@@ -93,8 +98,7 @@ class DashScopeEmbeddingConfig(BaseEmbeddingConfig):
         litellm_params: dict,
         stream: bool | None = None,
     ) -> str:
-        base = api_base or get_secret_str("DASHSCOPE_API_BASE") or DEFAULT_API_BASE
-        base = base.rstrip("/")
+        base: Final = self._resolve_embedding_api_base(api_base).rstrip("/")
         if base.endswith("/embeddings"):
             return base
         return f"{base}/embeddings"

litellm/llms/dashscope/image_generation/transformation.py

@@ -91,6 +91,15 @@ class DashScopeImageGenerationConfig(BaseImageGenerationConfig):
                 mapped[k] = v
         return mapped
 
+    def _resolve_api_key(self, api_key: str | None) -> str:
+        resolved_api_key: Final = api_key or get_secret_str("DASHSCOPE_API_KEY")
+        if not resolved_api_key:
+            raise ValueError("DASHSCOPE_API_KEY is not set")
+        return resolved_api_key
+
+    def _resolve_image_api_base(self, image_api_base: str | None) -> str:
+        return image_api_base or get_secret_str("DASHSCOPE_API_BASE_IMAGE") or DEFAULT_API_BASE
+
     def get_complete_url(
         self,
         api_base: str | None,
@@ -103,7 +112,7 @@ class DashScopeImageGenerationConfig(BaseImageGenerationConfig):
         image_api_base: Final = (
             api_base if api_base and not api_base.rstrip("/").endswith(CHAT_COMPATIBLE_MODE_PATH) else None
         )
-        return image_api_base or get_secret_str("DASHSCOPE_API_BASE_IMAGE") or DEFAULT_API_BASE
+        return self._resolve_image_api_base(image_api_base)
 
     def validate_environment(
         self,
@@ -115,10 +124,7 @@ class DashScopeImageGenerationConfig(BaseImageGenerationConfig):
         api_key: str | None = None,
         api_base: str | None = None,
     ) -> dict:
-        final_api_key: Final = api_key or get_secret_str("DASHSCOPE_API_KEY")
-        if not final_api_key:
-            raise ValueError("DASHSCOPE_API_KEY is not set")
-        headers["Authorization"] = f"Bearer {final_api_key}"
+        headers["Authorization"] = f"Bearer {self._resolve_api_key(api_key)}"
         headers["Content-Type"] = "application/json"
         return headers
 

litellm/llms/dashscope/qwen_ai_platform.py

@@ -0,0 +1,62 @@
+from typing import Final
+
+from litellm.secret_managers.main import get_secret_str
+
+from .chat.transformation import DashScopeChatConfig
+from .embed.transformation import DashScopeEmbeddingConfig
+from .image_generation.transformation import DashScopeImageGenerationConfig
+from .rerank.transformation import DashScopeRerankConfig
+
+QWEN_AI_PLATFORM_API_BASE: Final = "https://dashscope.aliyuncs.com/compatible-mode/v1"
+QWEN_AI_PLATFORM_RERANK_API_BASE: Final = "https://dashscope.aliyuncs.com/compatible-api/v1/reranks"
+QWEN_AI_PLATFORM_IMAGE_API_BASE: Final = (
+    "https://dashscope.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation"
+)
+
+
+def _resolve_qwen_ai_platform_api_key(api_key: str | None) -> str | None:
+    return api_key or get_secret_str("QWEN_AI_PLATFORM_API_KEY") or get_secret_str("DASHSCOPE_API_KEY")
+
+
+def _require_qwen_ai_platform_api_key(api_key: str | None) -> str:
+    resolved: Final = _resolve_qwen_ai_platform_api_key(api_key)
+    if resolved is None:
+        raise ValueError(
+            "Qwen AI Platform API key is required. Set 'QWEN_AI_PLATFORM_API_KEY' or 'DASHSCOPE_API_KEY' env var "
+            "or pass api_key explicitly."
+        )
+    return resolved
+
+
+class QwenAIPlatformChatConfig(DashScopeChatConfig):
+    def _get_openai_compatible_provider_info(
+        self, api_base: str | None, api_key: str | None
+    ) -> tuple[str | None, str | None]:
+        return self._resolve_chat_api_base(api_base), _resolve_qwen_ai_platform_api_key(api_key)
+
+    def _resolve_chat_api_base(self, api_base: str | None) -> str:
+        return api_base or get_secret_str("QWEN_AI_PLATFORM_API_BASE") or QWEN_AI_PLATFORM_API_BASE
+
+
+class QwenAIPlatformEmbeddingConfig(DashScopeEmbeddingConfig):
+    def _resolve_api_key(self, api_key: str | None) -> str:
+        return _require_qwen_ai_platform_api_key(api_key)
+
+    def _resolve_embedding_api_base(self, api_base: str | None) -> str:
+        return api_base or get_secret_str("QWEN_AI_PLATFORM_API_BASE") or QWEN_AI_PLATFORM_API_BASE
+
+
+class QwenAIPlatformRerankConfig(DashScopeRerankConfig):
+    def _resolve_api_key(self, api_key: str | None) -> str:
+        return _require_qwen_ai_platform_api_key(api_key)
+
+    def _resolve_rerank_api_base(self, api_base: str | None) -> str:
+        return api_base or get_secret_str("QWEN_AI_PLATFORM_API_BASE_RERANK") or QWEN_AI_PLATFORM_RERANK_API_BASE
+
+
+class QwenAIPlatformImageGenerationConfig(DashScopeImageGenerationConfig):
+    def _resolve_api_key(self, api_key: str | None) -> str:
+        return _require_qwen_ai_platform_api_key(api_key)
+
+    def _resolve_image_api_base(self, image_api_base: str | None) -> str:
+        return image_api_base or get_secret_str("QWEN_AI_PLATFORM_API_BASE_IMAGE") or QWEN_AI_PLATFORM_IMAGE_API_BASE

litellm/llms/dashscope/qwencloud.py

@@ -0,0 +1,62 @@
+from typing import Final
+
+from litellm.secret_managers.main import get_secret_str
+
+from .chat.transformation import DashScopeChatConfig
+from .embed.transformation import DashScopeEmbeddingConfig
+from .image_generation.transformation import DashScopeImageGenerationConfig
+from .rerank.transformation import DashScopeRerankConfig
+
+QWENCLOUD_API_BASE: Final = "https://dashscope-intl.aliyuncs.com/compatible-mode/v1"
+QWENCLOUD_RERANK_API_BASE: Final = "https://dashscope-intl.aliyuncs.com/compatible-api/v1/reranks"
+QWENCLOUD_IMAGE_API_BASE: Final = (
+    "https://dashscope-intl.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation"
+)
+
+
+def _resolve_qwencloud_api_key(api_key: str | None) -> str | None:
+    return api_key or get_secret_str("QWENCLOUD_API_KEY") or get_secret_str("DASHSCOPE_API_KEY")
+
+
+def _require_qwencloud_api_key(api_key: str | None) -> str:
+    resolved: Final = _resolve_qwencloud_api_key(api_key)
+    if resolved is None:
+        raise ValueError(
+            "QwenCloud API key is required. Set 'QWENCLOUD_API_KEY' or 'DASHSCOPE_API_KEY' env var "
+            "or pass api_key explicitly."
+        )
+    return resolved
+
+
+class QwenCloudChatConfig(DashScopeChatConfig):
+    def _get_openai_compatible_provider_info(
+        self, api_base: str | None, api_key: str | None
+    ) -> tuple[str | None, str | None]:
+        return self._resolve_chat_api_base(api_base), _resolve_qwencloud_api_key(api_key)
+
+    def _resolve_chat_api_base(self, api_base: str | None) -> str:
+        return api_base or get_secret_str("QWENCLOUD_API_BASE") or QWENCLOUD_API_BASE
+
+
+class QwenCloudEmbeddingConfig(DashScopeEmbeddingConfig):
+    def _resolve_api_key(self, api_key: str | None) -> str:
+        return _require_qwencloud_api_key(api_key)
+
+    def _resolve_embedding_api_base(self, api_base: str | None) -> str:
+        return api_base or get_secret_str("QWENCLOUD_API_BASE") or QWENCLOUD_API_BASE
+
+
+class QwenCloudRerankConfig(DashScopeRerankConfig):
+    def _resolve_api_key(self, api_key: str | None) -> str:
+        return _require_qwencloud_api_key(api_key)
+
+    def _resolve_rerank_api_base(self, api_base: str | None) -> str:
+        return api_base or get_secret_str("QWENCLOUD_API_BASE_RERANK") or QWENCLOUD_RERANK_API_BASE
+
+
+class QwenCloudImageGenerationConfig(DashScopeImageGenerationConfig):
+    def _resolve_api_key(self, api_key: str | None) -> str:
+        return _require_qwencloud_api_key(api_key)
+
+    def _resolve_image_api_base(self, image_api_base: str | None) -> str:
+        return image_api_base or get_secret_str("QWENCLOUD_API_BASE_IMAGE") or QWENCLOUD_IMAGE_API_BASE

litellm/llms/dashscope/rerank/transformation.py

@@ -58,19 +58,30 @@ class DashScopeRerankConfig(BaseRerankConfig):
     def __init__(self) -> None:
         pass
 
+    def _resolve_api_key(self, api_key: str | None) -> str:
+        resolved_api_key: Final = api_key if api_key is not None else get_secret_str("DASHSCOPE_API_KEY")
+        if resolved_api_key is None:
+            raise ValueError(
+                "DashScope API key is required. Set 'DASHSCOPE_API_KEY' env var or pass api_key explicitly."
+            )
+        return resolved_api_key
+
+    def _resolve_rerank_api_base(self, api_base: str | None) -> str:
+        if api_base is not None:
+            return api_base
+        return get_secret_str("DASHSCOPE_API_BASE_RERANK") or DEFAULT_RERANK_URL
+
     def get_complete_url(
         self,
         api_base: str | None,
         model: str,
         optional_params: dict | None = None,
     ) -> str:
-        if api_base is None:
-            api_base = get_secret_str("DASHSCOPE_API_BASE_RERANK") or DEFAULT_RERANK_URL
+        resolved_api_base: Final = self._resolve_rerank_api_base(api_base)
+        if resolved_api_base == DEFAULT_RERANK_URL:
+            return resolved_api_base
 
-        if api_base == DEFAULT_RERANK_URL:
-            return DEFAULT_RERANK_URL
-
-        cleaned: Final = api_base.rstrip("/")
+        cleaned: Final = resolved_api_base.rstrip("/")
         if cleaned.endswith("/reranks") or cleaned.endswith("/rerank"):
             return cleaned
 
@@ -88,19 +99,12 @@ class DashScopeRerankConfig(BaseRerankConfig):
         optional_params: dict | None = None,
         litellm_params: Mapping[str, object] | None = None,
     ) -> dict:
-        if api_key is None:
-            api_key = get_secret_str("DASHSCOPE_API_KEY")
-        if api_key is None:
-            raise ValueError(
-                "DashScope API key is required. Set 'DASHSCOPE_API_KEY' env var or pass api_key explicitly."
-            )
-
-        default_headers: Final = {
-            "Authorization": f"Bearer {api_key}",
+        return {
+            "Authorization": f"Bearer {self._resolve_api_key(api_key)}",
             "accept": "application/json",
             "content-type": "application/json",
+            **headers,
         }
-        return {**default_headers, **headers}
 
     def get_supported_cohere_rerank_params(self, model: str) -> list:
         return ["query", "documents", "top_n", "return_documents"]

litellm/main.py

@@ -6949,12 +6949,18 @@ def embedding(
                 aembedding=aembedding,
                 headers=headers,
             )
-        elif custom_llm_provider == "dashscope":
-            dashscope_key: Final = api_key or litellm.api_key or get_secret_str("DASHSCOPE_API_KEY")
+        elif custom_llm_provider in ("dashscope", "qwencloud", "qwen_ai_platform"):
+            from litellm.llms.dashscope.common_utils import (
+                missing_dashscope_family_key_message,
+                resolve_dashscope_family_api_key,
+            )
+
+            dashscope_key: Final = resolve_dashscope_family_api_key(
+                custom_llm_provider=custom_llm_provider,
+                api_key=api_key or litellm.api_key,
+            )
             if dashscope_key is None:
-                raise ValueError(
-                    "Missing API key for DashScope. Set DASHSCOPE_API_KEY environment variable or pass api_key parameter."
-                )
+                raise ValueError(missing_dashscope_family_key_message(custom_llm_provider))
             if extra_headers is not None and isinstance(extra_headers, dict):
                 headers = extra_headers
             else:

litellm/provider_endpoints_support_backup.json

@@ -671,6 +671,42 @@
         "interactions": true
       }
     },
+    "qwencloud": {
+      "display_name": "QwenCloud (`qwencloud`)",
+      "url": "https://docs.litellm.ai/docs/providers/qwencloud",
+      "endpoints": {
+        "chat_completions": true,
+        "messages": true,
+        "responses": true,
+        "embeddings": true,
+        "image_generations": true,
+        "audio_transcriptions": false,
+        "audio_speech": false,
+        "moderations": false,
+        "batches": false,
+        "rerank": true,
+        "a2a": true,
+        "interactions": true
+      }
+    },
+    "qwen_ai_platform": {
+      "display_name": "Qwen AI Platform (`qwen_ai_platform`)",
+      "url": "https://docs.litellm.ai/docs/providers/qwencloud",
+      "endpoints": {
+        "chat_completions": true,
+        "messages": true,
+        "responses": true,
+        "embeddings": true,
+        "image_generations": true,
+        "audio_transcriptions": false,
+        "audio_speech": false,
+        "moderations": false,
+        "batches": false,
+        "rerank": true,
+        "a2a": true,
+        "interactions": true
+      }
+    },
     "databricks": {
       "display_name": "Databricks (`databricks`)",
       "url": "https://docs.litellm.ai/docs/providers/databricks",

litellm/proxy/proxy_cli.py

@@ -913,13 +913,14 @@ class ProxyInitializationHelpers:
     envvar="ENFORCE_PRISMA_MIGRATION_CHECK",
 )
 @click.option(
-    "--use_v2_migration_resolver",
-    is_flag=True,
-    default=False,
+    "--use_v2_migration_resolver/--use_legacy_migration_resolver",
+    default=True,
     help=(
-        "Opt into the v2 migration resolver. Avoids the diff-and-force recovery "
-        "path that can cause schema thrashing during rolling deploys where two "
-        "LiteLLM versions contend for the same DB. Default is the v1 resolver."
+        "Which database migration resolver to run at startup. The default v2 "
+        "resolver avoids the diff-and-force recovery path that can cause schema "
+        "thrashing during rolling deploys where two LiteLLM versions contend for "
+        "the same DB. Pass --use_legacy_migration_resolver, or set "
+        "USE_V2_MIGRATION_RESOLVER=false, to fall back to v1."
     ),
     envvar="USE_V2_MIGRATION_RESOLVER",
 )
@@ -1310,10 +1311,11 @@ def run_server(
                 else:
                     if not use_v2_migration_resolver:
                         print(
-                            "\033[1;33mLiteLLM Proxy: Using default (v1) migration resolver. "
-                            "If your deployment has seen schema thrashing during rolling "
-                            "deploys, try --use_v2_migration_resolver (safer: avoids the "
-                            "diff-and-force recovery that caused the thrash).\033[0m"
+                            "\033[1;33mLiteLLM Proxy: Using the legacy (v1) migration resolver. "
+                            "The default v2 resolver is safer: it avoids the diff-and-force "
+                            "recovery that caused schema thrashing during rolling deploys. "
+                            "Remove --use_legacy_migration_resolver / "
+                            "USE_V2_MIGRATION_RESOLVER=false to switch back to it.\033[0m"
                         )
                     try:
                         setup_ok: Final = PrismaManager.setup_database(
@@ -1321,10 +1323,10 @@ def run_server(
                             use_v2_resolver=use_v2_migration_resolver,
                         )
                     except RuntimeError as e:
-                        # Raised on unrecoverable migration errors: the v2
-                        # resolver's non-idempotent failures and permission
-                        # issues, and any `prisma db push` against a
-                        # partitioned LiteLLM_SpendLogs.
+                        # Raised on unrecoverable migration errors: permission
+                        # failures from either resolver, the v2 resolver's
+                        # non-idempotent failures, and any `prisma db push`
+                        # against a partitioned LiteLLM_SpendLogs.
                         print(
                             f"\033[1;31mLiteLLM Proxy: Database migration cannot proceed. {e}\033[0m",
                             file=sys.stderr,

litellm/proxy/public_endpoints/provider_create_fields.json

@@ -986,6 +986,62 @@
     ],
     "default_model_placeholder": "gpt-3.5-turbo"
   },
+  {
+    "provider": "QwenCloud",
+    "provider_display_name": "QwenCloud",
+    "litellm_provider": "qwencloud",
+    "credential_fields": [
+      {
+        "key": "api_key",
+        "label": "QwenCloud API Key",
+        "placeholder": null,
+        "tooltip": null,
+        "required": true,
+        "field_type": "password",
+        "options": null,
+        "default_value": null
+      },
+      {
+        "key": "api_base",
+        "label": "API Base",
+        "placeholder": "https://dashscope-intl.aliyuncs.com/compatible-mode/v1",
+        "tooltip": "The base URL for QwenCloud. Defaults to https://dashscope-intl.aliyuncs.com/compatible-mode/v1 if not specified.",
+        "required": true,
+        "field_type": "text",
+        "options": null,
+        "default_value": "https://dashscope-intl.aliyuncs.com/compatible-mode/v1"
+      }
+    ],
+    "default_model_placeholder": "gpt-3.5-turbo"
+  },
+  {
+    "provider": "Qwen_AI_Platform",
+    "provider_display_name": "Qwen AI Platform",
+    "litellm_provider": "qwen_ai_platform",
+    "credential_fields": [
+      {
+        "key": "api_key",
+        "label": "Qwen AI Platform API Key",
+        "placeholder": null,
+        "tooltip": null,
+        "required": true,
+        "field_type": "password",
+        "options": null,
+        "default_value": null
+      },
+      {
+        "key": "api_base",
+        "label": "API Base",
+        "placeholder": "https://dashscope.aliyuncs.com/compatible-mode/v1",
+        "tooltip": "The base URL for Qwen AI Platform. Defaults to https://dashscope.aliyuncs.com/compatible-mode/v1 if not specified.",
+        "required": true,
+        "field_type": "text",
+        "options": null,
+        "default_value": "https://dashscope.aliyuncs.com/compatible-mode/v1"
+      }
+    ],
+    "default_model_placeholder": "gpt-3.5-turbo"
+  },
   {
     "provider": "Databricks",
     "provider_display_name": "Databricks",

litellm/types/utils.py

@@ -3771,6 +3771,8 @@ class LlmProviders(str, Enum):
     CODESTRAL = "codestral"
     TEXT_COMPLETION_CODESTRAL = "text-completion-codestral"
     DASHSCOPE = "dashscope"
+    QWENCLOUD = "qwencloud"
+    QWEN_AI_PLATFORM = "qwen_ai_platform"
     MODELSCOPE = "modelscope"
     MOONSHOT = "moonshot"
     PUBLICAI = "publicai"

litellm/utils.py

@@ -6586,11 +6586,11 @@ def validate_environment(
                 keys_in_environment = True
             else:
                 missing_keys.append("WANDB_API_KEY")
-        elif custom_llm_provider == "dashscope":
-            if "DASHSCOPE_API_KEY" in os.environ:
+        elif custom_llm_provider in ("dashscope", "qwencloud", "qwen_ai_platform"):
+            if f"{custom_llm_provider.upper()}_API_KEY" in os.environ or "DASHSCOPE_API_KEY" in os.environ:
                 keys_in_environment = True
             else:
-                missing_keys.append("DASHSCOPE_API_KEY")
+                missing_keys.append(f"{custom_llm_provider.upper()}_API_KEY")
         elif custom_llm_provider == "modelscope":
             if "MODELSCOPE_API_KEY" in os.environ:
                 keys_in_environment = True
@@ -8152,6 +8152,11 @@ class ProviderConfigManager:
             LlmProviders.NEBIUS: (lambda: litellm.NebiusConfig(), False),
             LlmProviders.WANDB: (lambda: litellm.WandbConfig(), False),
             LlmProviders.DASHSCOPE: (lambda: litellm.DashScopeChatConfig(), False),
+            LlmProviders.QWENCLOUD: (lambda: litellm.QwenCloudChatConfig(), False),
+            LlmProviders.QWEN_AI_PLATFORM: (
+                lambda: litellm.QwenAIPlatformChatConfig(),
+                False,
+            ),
             LlmProviders.MODELSCOPE: (lambda: litellm.ModelScopeChatConfig(), False),
             LlmProviders.MOONSHOT: (lambda: litellm.MoonshotChatConfig(), False),
             LlmProviders.DOCKER_MODEL_RUNNER: (
@@ -8366,12 +8371,16 @@ class ProviderConfigManager:
             )
 
             return VolcEngineEmbeddingConfig()
-        elif litellm.LlmProviders.DASHSCOPE == provider:
-            from litellm.llms.dashscope.embed.transformation import (
-                DashScopeEmbeddingConfig,
+        elif provider in (
+            litellm.LlmProviders.DASHSCOPE,
+            litellm.LlmProviders.QWENCLOUD,
+            litellm.LlmProviders.QWEN_AI_PLATFORM,
+        ):
+            from litellm.llms.dashscope.common_utils import (
+                get_dashscope_family_embedding_config,
             )
 
-            return DashScopeEmbeddingConfig()
+            return get_dashscope_family_embedding_config(provider.value)
         elif litellm.LlmProviders.OVHCLOUD == provider:
             return litellm.OVHCloudEmbeddingConfig()
         elif litellm.LlmProviders.SNOWFLAKE == provider:
@@ -8444,12 +8453,16 @@ class ProviderConfigManager:
             return litellm.VoyageRerankConfig()
         elif litellm.LlmProviders.WATSONX == provider:
             return litellm.IBMWatsonXRerankConfig()
-        elif litellm.LlmProviders.DASHSCOPE == provider:
-            from litellm.llms.dashscope.rerank.transformation import (
-                DashScopeRerankConfig,
+        elif provider in (
+            litellm.LlmProviders.DASHSCOPE,
+            litellm.LlmProviders.QWENCLOUD,
+            litellm.LlmProviders.QWEN_AI_PLATFORM,
+        ):
+            from litellm.llms.dashscope.common_utils import (
+                get_dashscope_family_rerank_config,
             )
 
-            return DashScopeRerankConfig()
+            return get_dashscope_family_rerank_config(provider.value)
         return litellm.CohereRerankConfig()
 
     @staticmethod
@@ -9122,12 +9135,16 @@ class ProviderConfigManager:
             )
 
             return get_openrouter_image_generation_config(model)
-        elif LlmProviders.DASHSCOPE == provider:
-            from litellm.llms.dashscope.image_generation import (
-                get_dashscope_image_generation_config,
+        elif provider in (
+            LlmProviders.DASHSCOPE,
+            LlmProviders.QWENCLOUD,
+            LlmProviders.QWEN_AI_PLATFORM,
+        ):
+            from litellm.llms.dashscope.common_utils import (
+                get_dashscope_family_image_generation_config,
             )
 
-            return get_dashscope_image_generation_config(model)
+            return get_dashscope_family_image_generation_config(provider.value)
         elif LlmProviders.MODELSCOPE == provider:
             from litellm.llms.modelscope.image_generation import (
                 get_modelscope_image_generation_config,

provider_endpoints_support.json

@@ -724,6 +724,42 @@
         "interactions": true
       }
     },
+    "qwencloud": {
+      "display_name": "QwenCloud (`qwencloud`)",
+      "url": "https://docs.litellm.ai/docs/providers/qwencloud",
+      "endpoints": {
+        "chat_completions": true,
+        "messages": true,
+        "responses": true,
+        "embeddings": true,
+        "image_generations": true,
+        "audio_transcriptions": false,
+        "audio_speech": false,
+        "moderations": false,
+        "batches": false,
+        "rerank": true,
+        "a2a": true,
+        "interactions": true
+      }
+    },
+    "qwen_ai_platform": {
+      "display_name": "Qwen AI Platform (`qwen_ai_platform`)",
+      "url": "https://docs.litellm.ai/docs/providers/qwencloud",
+      "endpoints": {
+        "chat_completions": true,
+        "messages": true,
+        "responses": true,
+        "embeddings": true,
+        "image_generations": true,
+        "audio_transcriptions": false,
+        "audio_speech": false,
+        "moderations": false,
+        "batches": false,
+        "rerank": true,
+        "a2a": true,
+        "interactions": true
+      }
+    },
     "databricks": {
       "display_name": "Databricks (`databricks`)",
       "url": "https://docs.litellm.ai/docs/providers/databricks",

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "litellm"
-version = "1.100.0"
+version = "1.101.0"
 description = "Library to easily interface with LLM API providers"
 readme = "README.md"
 requires-python = ">=3.10, <3.15"
@@ -67,8 +67,8 @@ proxy = [
     "azure-identity>=1.25.2,<2.0",
     "azure-storage-blob>=12.28.0,<13.0",
     "mcp>=1.28.1,<2.0",
-    "litellm-proxy-extras==0.4.91",
-    "litellm-enterprise==0.1.62",
+    "litellm-proxy-extras==0.4.92",
+    "litellm-enterprise==0.1.63",
     "RestrictedPython>=8.5,<9.0",
     "rich>=13.9.4,<14.0",
     "InquirerPy>=0.3.4,<1.0",
@@ -319,7 +319,7 @@ members = ["enterprise", "litellm-proxy-extras"]
 profile = "black"
 
 [tool.commitizen]
-version = "1.100.0"
+version = "1.101.0"
 version_files = [
     "pyproject.toml:^version",
 ]

tests/e2e/ui/tests/budgets/budgets.spec.ts

@@ -0,0 +1,133 @@
+import { test, expect, type Page as PlaywrightPage } from "@playwright/test";
+import { ADMIN_STORAGE_PATH } from "../../constants";
+import { Page } from "../../fixtures/pages";
+import { navigateToPage, dismissFeedbackPopup } from "../../helpers/navigation";
+import { masterKey } from "../../helpers/traffic";
+
+interface StoredBudget {
+  budget_id: string;
+  max_budget: number | null;
+  tpm_limit: number | null;
+  rpm_limit: number | null;
+  budget_duration: string | null;
+}
+
+/** A different route from the one the table renders from, so a row that only lives in its cache fails here. */
+async function findBudget(page: PlaywrightPage, budgetId: string): Promise<StoredBudget | undefined> {
+  const res = await page.request.get("/budget/list", {
+    headers: { Authorization: `Bearer ${masterKey()}` },
+  });
+  expect(res.ok(), `GET /budget/list (${res.status()})`).toBe(true);
+  return ((await res.json()) as StoredBudget[]).find((row) => row.budget_id === budgetId);
+}
+
+async function createBudgetViaApi(page: PlaywrightPage, budget: Partial<StoredBudget>): Promise<void> {
+  const res = await page.request.post("/budget/new", {
+    headers: { Authorization: `Bearer ${masterKey()}` },
+    data: budget,
+  });
+  expect(res.ok(), `POST /budget/new failed (${res.status()}): ${await res.text()}`).toBe(true);
+}
+
+async function searchForBudget(page: PlaywrightPage, budgetId: string): Promise<void> {
+  await page.getByPlaceholder("Search by budget ID").fill(budgetId);
+}
+
+test.describe("Budgets", () => {
+  test.use({ storageState: ADMIN_STORAGE_PATH });
+
+  test("Create a budget with rate limits and a spend cap", async ({ page }) => {
+    const budgetId = `e2e-budget-create-${Date.now()}`;
+
+    await navigateToPage(page, Page.Budgets);
+    await dismissFeedbackPopup(page);
+
+    await page.getByRole("button", { name: "Create Budget" }).click();
+
+    const modal = page.getByRole("dialog", { name: "Create Budget" });
+    await expect(modal).toBeVisible({ timeout: 10_000 });
+
+    await modal.getByRole("textbox", { name: "Budget ID" }).fill(budgetId);
+    await modal.getByRole("spinbutton", { name: "Max Tokens per minute" }).fill("5000");
+    await modal.getByRole("spinbutton", { name: "Max Requests per minute" }).fill("60");
+
+    await modal.getByRole("button", { name: "Optional Settings" }).click();
+    await modal.getByRole("spinbutton", { name: "Max Budget (USD)" }).fill("25.5");
+    await modal.getByRole("combobox", { name: "Reset Budget" }).click();
+    await page.getByRole("option", { name: "weekly" }).click();
+
+    await modal.getByRole("button", { name: "Create Budget" }).click();
+    await expect(modal).not.toBeVisible({ timeout: 10_000 });
+
+    await searchForBudget(page, budgetId);
+    const row = page.getByRole("row").filter({ hasText: budgetId });
+    await expect(row).toBeVisible({ timeout: 10_000 });
+    await expect(row).toContainText("$25.50");
+
+    const stored = await findBudget(page, budgetId);
+    expect(stored, `budget ${budgetId} readable from /budget/list`).toBeTruthy();
+    expect(stored?.max_budget, "spend cap persisted").toBe(25.5);
+    expect(stored?.tpm_limit, "TPM limit persisted").toBe(5000);
+    expect(stored?.rpm_limit, "RPM limit persisted").toBe(60);
+    expect(stored?.budget_duration, "reset window persisted").toBe("7d");
+  });
+
+  test("Raising a budget's spend cap leaves its rate limits alone", async ({ page }) => {
+    const budgetId = `e2e-budget-edit-${Date.now()}`;
+    await createBudgetViaApi(page, { budget_id: budgetId, max_budget: 10, tpm_limit: 1000, rpm_limit: 20 });
+
+    await navigateToPage(page, Page.Budgets);
+    await dismissFeedbackPopup(page);
+
+    await searchForBudget(page, budgetId);
+    await expect(page.getByRole("row").filter({ hasText: budgetId })).toBeVisible({ timeout: 10_000 });
+
+    await page.getByTestId(`budget-actions-${budgetId}`).click();
+    await page.getByTestId("budget-action-edit").click();
+
+    const modal = page.getByRole("dialog", { name: "Edit Budget" });
+    await expect(modal).toBeVisible({ timeout: 10_000 });
+
+    await modal.getByRole("button", { name: "Optional Settings" }).click();
+    await modal.getByRole("spinbutton", { name: "Max Budget (USD)" }).fill("99");
+    await modal.getByRole("button", { name: "Save", exact: true }).click();
+    await expect(modal).not.toBeVisible({ timeout: 10_000 });
+
+    await expect(page.getByRole("row").filter({ hasText: budgetId })).toContainText("$99.00", { timeout: 10_000 });
+
+    // Not hypothetical: the edit form posts the whole budget, so a field it fails to
+    // seed from the existing row goes to the server as null and silently clears.
+    const stored = await findBudget(page, budgetId);
+    expect(stored?.max_budget, "spend cap raised").toBe(99);
+    expect(stored?.tpm_limit, "TPM limit untouched by a spend-cap edit").toBe(1000);
+    expect(stored?.rpm_limit, "RPM limit untouched by a spend-cap edit").toBe(20);
+  });
+
+  test("Delete a budget", async ({ page }) => {
+    const budgetId = `e2e-budget-delete-${Date.now()}`;
+    await createBudgetViaApi(page, { budget_id: budgetId, max_budget: 5 });
+
+    await navigateToPage(page, Page.Budgets);
+    await dismissFeedbackPopup(page);
+
+    await searchForBudget(page, budgetId);
+    await expect(page.getByRole("row").filter({ hasText: budgetId })).toBeVisible({ timeout: 10_000 });
+
+    await page.getByTestId(`budget-actions-${budgetId}`).click();
+    await page.getByTestId("budget-action-delete").click();
+
+    const modal = page.getByRole("dialog", { name: "Delete Budget?" });
+    await expect(modal).toBeVisible({ timeout: 5_000 });
+    await modal.getByRole("button", { name: "Delete", exact: true }).click();
+
+    await expect(page.getByRole("row").filter({ hasText: budgetId })).toHaveCount(0, { timeout: 10_000 });
+
+    // The row disappearing is a cache invalidation; the budget is gone when the route stops serving it.
+    await expect
+      .poll(async () => await findBudget(page, budgetId), {
+        message: `budget ${budgetId} still readable from /budget/list after delete`,
+        timeout: 15_000,
+      })
+      .toBeUndefined();
+  });
+});

tests/e2e/ui/tests/guardrails/guardrails.spec.ts

@@ -1,11 +1,213 @@
-import { test, expect } from "@playwright/test";
+import { test, expect, type Page as PlaywrightPage } from "@playwright/test";
 import { ADMIN_STORAGE_PATH, E2E_TEAM_NO_ADMIN_ID } from "../../constants";
 import { Page } from "../../fixtures/pages";
 import { navigateToPage, dismissFeedbackPopup, clickTeamId } from "../../helpers/navigation";
+import { CHAT_MODEL_A, MOCK_RESPONSE_TEXT, masterKey } from "../../helpers/traffic";
+
+interface StoredGuardrail {
+  guardrail_id: string;
+  guardrail_name: string | null;
+}
+
+async function listGuardrails(page: PlaywrightPage): Promise<StoredGuardrail[]> {
+  const res = await page.request.get("/v2/guardrails/list", {
+    headers: { Authorization: `Bearer ${masterKey()}` },
+  });
+  expect(res.ok(), `GET /v2/guardrails/list (${res.status()})`).toBe(true);
+  return ((await res.json()) as { guardrails: StoredGuardrail[] }).guardrails;
+}
+
+async function findGuardrail(page: PlaywrightPage, name: string): Promise<StoredGuardrail | undefined> {
+  return (await listGuardrails(page)).find((row) => row.guardrail_name === name);
+}
+
+const createdGuardrails: string[] = [];
+
+async function createKeywordGuardrailViaApi(page: PlaywrightPage, name: string, keyword: string): Promise<string> {
+  const res = await page.request.post("/guardrails", {
+    headers: { Authorization: `Bearer ${masterKey()}` },
+    data: {
+      guardrail: {
+        guardrail_name: name,
+        litellm_params: {
+          guardrail: "litellm_content_filter",
+          mode: "pre_call",
+          default_on: false,
+          blocked_words: [{ keyword, action: "BLOCK" }],
+        },
+      },
+    },
+  });
+  expect(res.ok(), `POST /guardrails failed (${res.status()}): ${await res.text()}`).toBe(true);
+  createdGuardrails.push(name);
+  const guardrail = await findGuardrail(page, name);
+  expect(guardrail?.guardrail_id, `guardrail ${name} has an id`).toBeTruthy();
+  return guardrail!.guardrail_id;
+}
+
+async function openKeywordsStep(page: PlaywrightPage, name: string) {
+  await page.getByRole("button", { name: "Add New Guardrail" }).click();
+  await page.getByRole("menuitem", { name: "Add Provider Guardrail" }).click();
+
+  const wizard = page.getByRole("dialog", { name: "Create guardrail" });
+  await expect(wizard).toBeVisible({ timeout: 10_000 });
+
+  await wizard.getByRole("textbox", { name: "Guardrail Name" }).fill(name);
+  await wizard.getByRole("combobox", { name: "Guardrail Provider" }).click();
+  // The content filter runs inside the proxy, so this is the one provider a test can
+  // configure end to end without standing up a third-party moderation service.
+  await page.getByRole("option", { name: /LiteLLM Content Filter/ }).click();
+
+  for (const step of ["Topics", "Patterns", "Keywords"]) {
+    await wizard.getByRole("button", { name: "Next" }).click();
+    await expect(wizard).toContainText(step, { timeout: 10_000 });
+  }
+  return wizard;
+}
 
 test.describe("Guardrails", () => {
   test.use({ storageState: ADMIN_STORAGE_PATH });
 
+  test.afterEach(async ({ page }) => {
+    // Guardrails live in the database and show up in the table and the playground list, so a run
+    // that leaves them behind changes what the next run sees.
+    for (const name of createdGuardrails.splice(0)) {
+      const guardrail = await findGuardrail(page, name);
+      if (guardrail) {
+        const deleted = await page.request.delete(`/guardrails/${guardrail.guardrail_id}`, {
+          headers: { Authorization: `Bearer ${masterKey()}` },
+        });
+        expect(deleted.ok(), `DELETE /guardrails/${guardrail.guardrail_id} (${deleted.status()})`).toBe(true);
+      }
+    }
+  });
+
+  test("A guardrail created through the wizard blocks the keyword it was given", async ({ page }) => {
+    const stamp = Date.now();
+    const guardrailName = `e2e-guardrail-create-${stamp}`;
+    // Unique per run so a concurrent test's prompt can never trip this guardrail, or vice versa.
+    const bannedKeyword = `e2ebanned${stamp}`;
+
+    await navigateToPage(page, Page.Guardrails);
+    await dismissFeedbackPopup(page);
+
+    createdGuardrails.push(guardrailName);
+    const wizard = await openKeywordsStep(page, guardrailName);
+
+    await wizard.getByRole("button", { name: "Add keyword" }).click();
+    const keywordModal = page.getByRole("dialog", { name: "Add blocked keyword" });
+    await expect(keywordModal).toBeVisible({ timeout: 10_000 });
+    await keywordModal.getByPlaceholder("Enter sensitive keyword or phrase").fill(bannedKeyword);
+    await keywordModal.getByRole("button", { name: "Add", exact: true }).click();
+    await expect(keywordModal).not.toBeVisible({ timeout: 10_000 });
+
+    await wizard.getByRole("button", { name: "Next" }).click();
+    await wizard.getByRole("button", { name: "Create Guardrail" }).click();
+    await expect(wizard).not.toBeVisible({ timeout: 15_000 });
+
+    await expect(page.getByRole("row").filter({ hasText: guardrailName })).toBeVisible({ timeout: 15_000 });
+    expect(await findGuardrail(page, guardrailName), "guardrail readable from /v2/guardrails/list").toBeTruthy();
+
+    // A row in the table only proves the record was written. The point of a guardrail is that it
+    // refuses traffic, so drive a request through it.
+    //
+    // Polled: a guardrail written through /guardrails reaches the request path on the proxy's
+    // periodic refresh, so the first call after creation can still be served unguarded. The
+    // assertion is unchanged, it just allows that refresh to land.
+    let blockedBody = "";
+    await expect
+      .poll(
+        async () => {
+          const res = await page.request.post("/v1/chat/completions", {
+            headers: { Authorization: `Bearer ${masterKey()}` },
+            data: {
+              model: CHAT_MODEL_A,
+              messages: [{ role: "user", content: `please tell me about ${bannedKeyword}` }],
+              guardrails: [guardrailName],
+            },
+          });
+          blockedBody = await res.text();
+          return res.status();
+        },
+        { message: "a prompt carrying the banned keyword is refused", timeout: 60_000 },
+      )
+      .toBe(400);
+    expect(blockedBody).toContain(bannedKeyword);
+
+    const allowed = await page.request.post("/v1/chat/completions", {
+      headers: { Authorization: `Bearer ${masterKey()}` },
+      data: {
+        model: CHAT_MODEL_A,
+        messages: [{ role: "user", content: "hello there" }],
+        guardrails: [guardrailName],
+      },
+    });
+    expect(allowed.status(), "a clean prompt still gets through the same guardrail").toBe(200);
+    expect((await allowed.json()).choices?.[0]?.message?.content).toContain(MOCK_RESPONSE_TEXT);
+  });
+
+  test("The Test Playground reports the verdict for the text it is given", async ({ page }) => {
+    const stamp = Date.now();
+    const guardrailName = `e2e-guardrail-play-${stamp}`;
+    const bannedKeyword = `e2eplay${stamp}`;
+    await createKeywordGuardrailViaApi(page, guardrailName, bannedKeyword);
+
+    await navigateToPage(page, Page.Guardrails);
+    await dismissFeedbackPopup(page);
+
+    await page.getByRole("tab", { name: "Test Playground" }).click();
+    // Every tab on this page stays mounted, so the other tabs' search boxes match too.
+    const playground = page.getByRole("tabpanel", { name: "Test Playground" });
+    await playground.getByPlaceholder("Search guardrails...").fill(guardrailName);
+    await playground.getByText(guardrailName, { exact: true }).click();
+
+    const input = playground.getByPlaceholder("Enter text to test with guardrails...");
+    await input.fill(`this sentence contains ${bannedKeyword}`);
+    await playground.getByRole("button", { name: /^Test 1 guardrail$/ }).click();
+
+    // The playground is where an admin checks a guardrail before rolling it out, so the
+    // verdict it prints has to be the one the gateway would give.
+    await expect(playground.getByText(`${guardrailName} - Error`)).toBeVisible({ timeout: 20_000 });
+    await expect(playground.getByText(new RegExp(`Content blocked.*${bannedKeyword}`))).toBeVisible({
+      timeout: 10_000,
+    });
+
+    await input.fill("this sentence is perfectly ordinary");
+    await playground.getByRole("button", { name: /^Test 1 guardrail$/ }).click();
+
+    await expect(playground.getByText(`${guardrailName} - Error`)).toHaveCount(0, { timeout: 20_000 });
+    await expect(playground.getByText("this sentence is perfectly ordinary").last()).toBeVisible({ timeout: 10_000 });
+  });
+
+  test("Delete a guardrail", async ({ page }) => {
+    const stamp = Date.now();
+    const guardrailName = `e2e-guardrail-delete-${stamp}`;
+    const guardrailId = await createKeywordGuardrailViaApi(page, guardrailName, `e2edelete${stamp}`);
+
+    await navigateToPage(page, Page.Guardrails);
+    await dismissFeedbackPopup(page);
+
+    await expect(page.getByRole("row").filter({ hasText: guardrailName })).toBeVisible({ timeout: 15_000 });
+
+    await page.getByTestId(`guardrail-actions-${guardrailId}`).click();
+    await page.getByTestId("guardrail-action-delete").click();
+
+    const modal = page.getByRole("dialog");
+    await expect(modal).toBeVisible({ timeout: 5_000 });
+    await modal.getByRole("button", { name: "Delete", exact: true }).click();
+
+    await expect(page.getByRole("row").filter({ hasText: guardrailName })).toHaveCount(0, { timeout: 15_000 });
+
+    // The RC checklist deletes then reloads, because a row vanishing from the table has
+    // fooled us before; assert against the route the reload would read.
+    await expect
+      .poll(async () => await findGuardrail(page, guardrailName), {
+        message: `guardrail ${guardrailName} still listed after delete`,
+        timeout: 15_000,
+      })
+      .toBeUndefined();
+  });
+
   test("Create a Presidio guardrail, see it in team settings, and delete it", async ({ page }) => {
     const guardrailName = `e2e-presidio-${Date.now()}`;
 

tests/litellm-proxy-extras/test_setup_database_fail_fast.py

@@ -0,0 +1,571 @@
+"""Regression tests for ProxyExtrasDBManager's v2 migration resolver.
+
+v2 is the proxy CLI default; v1 stays reachable via the `use_v2_resolver`
+kwarg, which still defaults to False for direct callers.
+"""
+
+import subprocess
+from unittest.mock import patch
+
+import pytest
+
+from litellm_proxy_extras.utils import (
+    _PRISMA_ATTEMPTS,
+    ProxyExtrasDBManager,
+    _max_migration_timestamp,
+    _migration_timestamp,
+)
+
+
+def _fake_migrate_deploy_failure(returncode: int, stderr: str):
+    def _run(*args, **kwargs):
+        raise subprocess.CalledProcessError(
+            returncode=returncode,
+            cmd=args[0],
+            stderr=stderr,
+            output="",
+        )
+
+    return _run
+
+
+def test_v2_p3018_permission_error_raises_runtime_error(monkeypatch, tmp_path):
+    """v2: a permission failure during migrate deploy raises RuntimeError."""
+    monkeypatch.setenv("DATABASE_URL", "postgresql://u:p@localhost:9/x")
+    monkeypatch.setattr(
+        ProxyExtrasDBManager, "_warn_if_db_ahead_of_head", lambda _: None
+    )
+    monkeypatch.setattr(ProxyExtrasDBManager, "_get_prisma_dir", lambda: str(tmp_path))
+    (tmp_path / "schema.prisma").write_text("// stub")
+
+    stderr = (
+        "Error: P3018\nMigration name: 20250326162113_baseline\n"
+        "Database error code: 42501\npermission denied for schema public"
+    )
+    with patch("subprocess.run", side_effect=_fake_migrate_deploy_failure(1, stderr)):
+        with pytest.raises(RuntimeError, match="permission"):
+            ProxyExtrasDBManager.setup_database(use_migrate=True, use_v2_resolver=True)
+
+
+def test_v2_non_idempotent_p3009_raises_runtime_error(monkeypatch, tmp_path):
+    """v2: a non-idempotent migration failure raises (no silent recovery)."""
+    monkeypatch.setenv("DATABASE_URL", "postgresql://u:p@localhost:9/x")
+    monkeypatch.setattr(
+        ProxyExtrasDBManager, "_warn_if_db_ahead_of_head", lambda _: None
+    )
+    monkeypatch.setattr(ProxyExtrasDBManager, "_get_prisma_dir", lambda: str(tmp_path))
+    (tmp_path / "schema.prisma").write_text("// stub")
+
+    stderr = (
+        "Error: P3009\nMigration `20260101000000_genuinely_broken` failed\n"
+        'Reason: syntax error at or near "BRKN" LINE 42'
+    )
+    with patch("subprocess.run", side_effect=_fake_migrate_deploy_failure(1, stderr)):
+        with pytest.raises(RuntimeError, match="cannot be auto-recovered"):
+            ProxyExtrasDBManager.setup_database(use_migrate=True, use_v2_resolver=True)
+
+
+def test_strip_prisma_query_params_removes_connection_limit():
+    """DATABASE_URLs with Prisma-specific params should be parseable by psycopg."""
+    url = "postgresql://u:p@h:5432/db?connection_limit=100&pool_timeout=60&sslmode=require"
+    stripped = ProxyExtrasDBManager._strip_prisma_query_params(url)
+    assert "connection_limit" not in stripped
+    assert "pool_timeout" not in stripped
+    assert "sslmode=require" in stripped
+
+
+def test_strip_prisma_query_params_passthrough_no_query():
+    """URLs without query strings are returned unchanged."""
+    url = "postgresql://u:p@h:5432/db"
+    assert ProxyExtrasDBManager._strip_prisma_query_params(url) == url
+
+
+def test_migration_timestamp_extracts_leading_digits():
+    assert _migration_timestamp("20260101000000_add_foo") == 20260101000000
+    assert _migration_timestamp("20250326162113_baseline") == 20250326162113
+
+
+def test_migration_timestamp_returns_zero_on_malformed():
+    assert _migration_timestamp("0_init") == 0
+    assert _migration_timestamp("not_a_migration") == 0
+
+
+def test_max_migration_timestamp():
+    names = {"20250326000000_a", "20260415000000_b", "20251115000000_c"}
+    assert _max_migration_timestamp(names) == 20260415000000
+
+
+def test_max_migration_timestamp_empty_set():
+    assert _max_migration_timestamp(set()) == 0
+
+
+def test_v1_default_still_calls_resolve_all_migrations(monkeypatch, tmp_path):
+    """v1 (default) continues to call _resolve_all_migrations on the happy path.
+
+    This is the existing buggy behavior — we're not fixing it in v1, only
+    offering v2 as opt-in. This test pins the default so that a future
+    inadvertent default flip is caught.
+    """
+    monkeypatch.setattr(ProxyExtrasDBManager, "_get_prisma_dir", lambda: str(tmp_path))
+    (tmp_path / "schema.prisma").write_text("// stub")
+
+    # Stub `prisma migrate deploy` to claim success with pending migrations
+    # applied, which is the code path that triggers the legacy post-migration
+    # sanity check (a call to _resolve_all_migrations).
+    class FakeResult:
+        stdout = "Applied migration.\n"
+        stderr = ""
+
+    def fake_run(cmd, *args, **kwargs):
+        return FakeResult()
+
+    resolve_called = {"n": 0}
+
+    def fake_resolve(*args, **kwargs):
+        resolve_called["n"] += 1
+
+    monkeypatch.setattr("subprocess.run", fake_run)
+    monkeypatch.setattr(ProxyExtrasDBManager, "_resolve_all_migrations", fake_resolve)
+
+    ok = ProxyExtrasDBManager.setup_database(use_migrate=True)  # v2 flag NOT set
+    assert ok is True
+    assert resolve_called["n"] == 1, "v1 default should still invoke the legacy path"
+
+
+def test_v2_db_push_wraps_subprocess_error_as_runtime_error(monkeypatch, tmp_path):
+    """v2: a failing `prisma db push` must raise RuntimeError, not leak
+    CalledProcessError past proxy_cli.py's `except RuntimeError`."""
+    monkeypatch.setattr(ProxyExtrasDBManager, "_get_prisma_dir", lambda: str(tmp_path))
+    (tmp_path / "schema.prisma").write_text("// stub")
+
+    stderr = "db push error"
+    with patch("subprocess.run", side_effect=_fake_migrate_deploy_failure(1, stderr)):
+        with pytest.raises(RuntimeError, match="prisma db push failed"):
+            ProxyExtrasDBManager.setup_database(use_migrate=False, use_v2_resolver=True)
+
+
+def test_v2_warn_ahead_of_head_swallows_db_errors(monkeypatch, tmp_path):
+    """_warn_if_db_ahead_of_head must never raise — it's informational.
+
+    Non-connection DB errors (e.g. InsufficientPrivilege from a user
+    without SELECT on _prisma_migrations) must be caught, not propagated.
+    """
+    import psycopg
+
+    monkeypatch.setenv("DATABASE_URL", "postgresql://u:p@localhost:9/x")
+    monkeypatch.setattr(ProxyExtrasDBManager, "_get_prisma_dir", lambda: str(tmp_path))
+    (tmp_path / "schema.prisma").write_text("// stub")
+
+    class _FakeConn:
+        def __enter__(self):
+            return self
+
+        def __exit__(self, *a):
+            return False
+
+        def execute(self, *a, **kw):
+            # Simulate an InsufficientPrivilege (subclass of DatabaseError).
+            raise psycopg.errors.InsufficientPrivilege("permission denied")
+
+    connects = {"n": 0}
+
+    def _fake_connect(*a, **kw):
+        connects["n"] += 1
+        return _FakeConn()
+
+    monkeypatch.setattr("psycopg.connect", _fake_connect)
+
+    assert ProxyExtrasDBManager._warn_if_db_ahead_of_head(str(tmp_path)) is None
+    assert connects["n"] == 1, "the failing query must actually have been reached"
+
+
+def test_v2_resolve_specific_migration_failure_raises_runtime_error(
+    monkeypatch, tmp_path
+):
+    """If marking a migration as applied fails inside P3009 idempotent
+    recovery, the subprocess error must be re-raised as RuntimeError so
+    proxy_cli.py catches it cleanly (instead of leaking CalledProcessError)."""
+    monkeypatch.setattr(
+        ProxyExtrasDBManager, "_warn_if_db_ahead_of_head", lambda _: None
+    )
+    monkeypatch.setattr(ProxyExtrasDBManager, "_get_prisma_dir", lambda: str(tmp_path))
+    (tmp_path / "schema.prisma").write_text("// stub")
+    monkeypatch.setattr(
+        ProxyExtrasDBManager, "_roll_back_migration", lambda *a, **kw: None
+    )
+
+    # First call: migrate deploy -> P3009 idempotent error.
+    # Recovery path tries _resolve_specific_migration; that also raises.
+    def _failing_resolve(*a, **kw):
+        raise subprocess.CalledProcessError(
+            returncode=1,
+            cmd="prisma migrate resolve --applied",
+            stderr="resolve failed",
+            output="",
+        )
+
+    monkeypatch.setattr(
+        ProxyExtrasDBManager, "_resolve_specific_migration", _failing_resolve
+    )
+
+    stderr = (
+        "Error: P3009\nMigration `20260101000000_some_migration` failed\n"
+        "relation already exists"
+    )
+    with patch("subprocess.run", side_effect=_fake_migrate_deploy_failure(1, stderr)):
+        with pytest.raises(
+            RuntimeError, match=r"Failed to mark migration .* as applied"
+        ):
+            ProxyExtrasDBManager.setup_database(use_migrate=True, use_v2_resolver=True)
+
+
+def test_v2_does_not_call_resolve_all_migrations(monkeypatch, tmp_path):
+    """v2 must never call _resolve_all_migrations — that's the bug it fixes."""
+    monkeypatch.setattr(
+        ProxyExtrasDBManager, "_warn_if_db_ahead_of_head", lambda _: None
+    )
+    monkeypatch.setattr(ProxyExtrasDBManager, "_get_prisma_dir", lambda: str(tmp_path))
+    (tmp_path / "schema.prisma").write_text("// stub")
+
+    class FakeResult:
+        stdout = "Applied migration.\n"
+        stderr = ""
+
+    monkeypatch.setattr("subprocess.run", lambda *a, **kw: FakeResult())
+
+    resolve_called = {"n": 0}
+    monkeypatch.setattr(
+        ProxyExtrasDBManager,
+        "_resolve_all_migrations",
+        lambda *a, **kw: resolve_called.__setitem__("n", resolve_called["n"] + 1),
+    )
+
+    ok = ProxyExtrasDBManager.setup_database(use_migrate=True, use_v2_resolver=True)
+    assert ok is True
+    assert resolve_called["n"] == 0, "v2 must not invoke the diff-and-force recovery"
+
+
+_DEADLOCK_STDERR = (
+    "Error: ERROR: deadlock detected\n"
+    "DETAIL: Process 277 waits for ExclusiveLock on advisory lock "
+    "[17556,0,72707369,1]; blocked by process 278.\n"
+    "Process 278 waits for ShareLock on virtual transaction 3/1041; "
+    "blocked by process 277."
+)
+
+
+class _DeployApplied:
+    stdout = "All migrations have been successfully applied."
+    stderr = ""
+    returncode = 0
+
+
+def _deploy_only(deploy_side_effect):
+    """subprocess.run stand-in that only intercepts `prisma migrate deploy`.
+
+    Scoped by argv so the Prisma toolchain check cannot consume the mock first.
+    """
+    deploys = {"n": 0}
+
+    def _run(*args, **kwargs):
+        cmd = args[0] if args else kwargs.get("args", [])
+        if list(cmd)[-2:] == ["migrate", "deploy"]:
+            deploys["n"] += 1
+            return deploy_side_effect(deploys["n"], cmd)
+        return _DeployApplied()
+
+    return _run, deploys
+
+
+def _prepare_v2_resolver(monkeypatch, tmp_path):
+    monkeypatch.setenv("DATABASE_URL", "postgresql://u:p@localhost:9/x")
+    monkeypatch.setattr(
+        ProxyExtrasDBManager, "_warn_if_db_ahead_of_head", lambda _: None
+    )
+    monkeypatch.setattr(ProxyExtrasDBManager, "_get_prisma_dir", lambda: str(tmp_path))
+    (tmp_path / "schema.prisma").write_text("// stub")
+    monkeypatch.setattr("time.sleep", lambda *_a, **_k: None)
+
+
+def test_v2_retries_transient_advisory_lock_deadlock(monkeypatch, tmp_path):
+    """v2: replicas racing `migrate deploy` deadlock on Prisma's advisory
+    lock, which is transient and must be retried rather than kill the boot."""
+    _prepare_v2_resolver(monkeypatch, tmp_path)
+
+    def _side_effect(n, cmd):
+        if n == 1:
+            raise subprocess.CalledProcessError(
+                returncode=1, cmd=cmd, stderr=_DEADLOCK_STDERR, output=""
+            )
+        return _DeployApplied()
+
+    run, deploys = _deploy_only(_side_effect)
+    with patch("subprocess.run", side_effect=run):
+        ok = ProxyExtrasDBManager.setup_database(use_migrate=True, use_v2_resolver=True)
+
+    assert ok is True
+    assert deploys["n"] == 2, "the deadlocked deploy must be retried, not raised"
+
+
+def test_v2_persistent_advisory_lock_deadlock_eventually_raises(monkeypatch, tmp_path):
+    """v2: the deadlock retry is bounded, so a deadlock that never clears
+    still raises instead of looping or reporting success."""
+    _prepare_v2_resolver(monkeypatch, tmp_path)
+
+    def _side_effect(n, cmd):
+        raise subprocess.CalledProcessError(
+            returncode=1, cmd=cmd, stderr=_DEADLOCK_STDERR, output=""
+        )
+
+    run, deploys = _deploy_only(_side_effect)
+    with patch("subprocess.run", side_effect=run):
+        with pytest.raises(RuntimeError, match="after 4 attempts"):
+            ProxyExtrasDBManager.setup_database(use_migrate=True, use_v2_resolver=True)
+
+    assert deploys["n"] == 4
+
+
+@pytest.mark.parametrize(
+    "stderr",
+    [
+        "Error: P1001: Can't reach database server at `db`:`5432`",
+        "Error: P1002: The database server was reached but timed out.",
+    ],
+)
+def test_v2_retries_transient_database_connectivity_errors(monkeypatch, tmp_path, stderr):
+    """v2: a database not accepting connections yet is retried, not fatal."""
+    _prepare_v2_resolver(monkeypatch, tmp_path)
+
+    def _side_effect(n, cmd):
+        if n == 1:
+            raise subprocess.CalledProcessError(
+                returncode=1, cmd=cmd, stderr=stderr, output=""
+            )
+        return _DeployApplied()
+
+    run, deploys = _deploy_only(_side_effect)
+    with patch("subprocess.run", side_effect=run):
+        ok = ProxyExtrasDBManager.setup_database(use_migrate=True, use_v2_resolver=True)
+
+    assert ok is True
+    assert deploys["n"] == 2, "an unreachable database must be retried, not raised"
+
+
+def test_v2_unreachable_database_still_fails_after_the_retries(monkeypatch, tmp_path):
+    """v2: a genuinely unreachable database still raises once the attempts
+    are spent, rather than passing as a successful migration."""
+    _prepare_v2_resolver(monkeypatch, tmp_path)
+
+    def _side_effect(n, cmd):
+        raise subprocess.CalledProcessError(
+            returncode=1,
+            cmd=cmd,
+            stderr="Error: P1001: Can't reach database server at `db`:`5432`",
+            output="",
+        )
+
+    run, deploys = _deploy_only(_side_effect)
+    with patch("subprocess.run", side_effect=run):
+        with pytest.raises(RuntimeError, match="after 4 attempts"):
+            ProxyExtrasDBManager.setup_database(use_migrate=True, use_v2_resolver=True)
+
+    assert deploys["n"] == 4
+
+
+def test_v2_exhausted_retries_report_the_prisma_error(monkeypatch, tmp_path, caplog):
+    """v2: retrying must not swallow Prisma's stderr, which is captured and is
+    the only place the cause appears for an operator or a boot-log grep."""
+    _prepare_v2_resolver(monkeypatch, tmp_path)
+    stderr = "Error: P1001: Can't reach database server at `wrong`:`5432`"
+
+    def _side_effect(n, cmd):
+        raise subprocess.CalledProcessError(
+            returncode=1, cmd=cmd, stderr=stderr, output=""
+        )
+
+    run, _ = _deploy_only(_side_effect)
+    with caplog.at_level("INFO", logger="litellm_proxy_extras"):
+        with patch("subprocess.run", side_effect=run):
+            with pytest.raises(RuntimeError) as exc_info:
+                ProxyExtrasDBManager.setup_database(
+                    use_migrate=True, use_v2_resolver=True
+                )
+
+    assert "P1001" in str(exc_info.value)
+    assert "P1001" in caplog.text
+
+
+def test_v2_db_push_retries_transient_failures(monkeypatch, tmp_path):
+    """v2: `prisma db push` retries a transient failure like v1 did.
+
+    Reached from the migrations Job (USE_PRISMA_DB_PUSH=true), not from the
+    proxy CLI, whose --use_prisma_db_push has its own loop in prisma_client.
+    """
+    _prepare_v2_resolver(monkeypatch, tmp_path)
+
+    pushes = {"n": 0}
+
+    def _run(*args, **kwargs):
+        cmd = list(args[0] if args else kwargs.get("args", []))
+        if cmd[-3:] != ["db", "push", "--accept-data-loss"]:
+            return _DeployApplied()
+        pushes["n"] += 1
+        if pushes["n"] == 1:
+            raise subprocess.CalledProcessError(
+                returncode=1,
+                cmd=cmd,
+                stderr="Error: P1001: Can't reach database server at `db`:`5432`",
+                output="",
+            )
+        return _DeployApplied()
+
+    monkeypatch.setattr(
+        ProxyExtrasDBManager, "spend_logs_is_partitioned", lambda: False
+    )
+    with patch("subprocess.run", side_effect=_run):
+        ok = ProxyExtrasDBManager.setup_database(use_migrate=False, use_v2_resolver=True)
+
+    assert ok is True
+    assert pushes["n"] == 2
+
+
+def test_v2_db_push_retries_are_bounded_and_report_the_prisma_error(
+    monkeypatch, tmp_path
+):
+    """v2: a database that never comes back stops after _PRISMA_ATTEMPTS and
+    surfaces the prisma error, rather than retrying the boot forever."""
+    _prepare_v2_resolver(monkeypatch, tmp_path)
+
+    pushes = {"n": 0}
+
+    def _run(*args, **kwargs):
+        cmd = list(args[0] if args else kwargs.get("args", []))
+        if cmd[-3:] != ["db", "push", "--accept-data-loss"]:
+            return _DeployApplied()
+        pushes["n"] += 1
+        raise subprocess.CalledProcessError(
+            returncode=1,
+            cmd=cmd,
+            stderr="Error: P1001: Can't reach database server at `db`:`5432`",
+            output="",
+        )
+
+    monkeypatch.setattr(
+        ProxyExtrasDBManager, "spend_logs_is_partitioned", lambda: False
+    )
+    with patch("subprocess.run", side_effect=_run):
+        with pytest.raises(RuntimeError) as exc:
+            ProxyExtrasDBManager.setup_database(
+                use_migrate=False, use_v2_resolver=True
+            )
+
+    assert pushes["n"] == _PRISMA_ATTEMPTS
+    assert "P1001" in str(exc.value)
+
+
+def _db_push_only(push_side_effect):
+    """subprocess.run stand-in that only intercepts `prisma db push`."""
+    pushes = {"n": 0}
+
+    def _run(*args, **kwargs):
+        cmd = list(args[0] if args else kwargs.get("args", []))
+        if cmd[-3:] != ["db", "push", "--accept-data-loss"]:
+            return _DeployApplied()
+        pushes["n"] += 1
+        return push_side_effect(pushes["n"], cmd)
+
+    return _run, pushes
+
+
+def _timed_out_for_real():
+    """Capture what subprocess.run really puts on a TimeoutExpired.
+
+    Under text=True it still leaves stderr as bytes, unlike CalledProcessError,
+    so hardcoding a str here would test a shape production never sees. Derived
+    at import, before any test patches subprocess.run.
+    """
+    try:
+        subprocess.run(
+            ["sh", "-c", "echo 'Error: P1001 unreachable' >&2; sleep 5"],
+            timeout=0.2,
+            check=True,
+            capture_output=True,
+            text=True,
+        )
+    except subprocess.TimeoutExpired as e:
+        return e
+    raise AssertionError("the helper command was supposed to time out")
+
+
+_TIMEOUT_TEMPLATE = _timed_out_for_real()
+
+
+def _real_timeout_expired(cmd):
+    return subprocess.TimeoutExpired(
+        cmd=cmd,
+        timeout=_TIMEOUT_TEMPLATE.timeout,
+        output=_TIMEOUT_TEMPLATE.stdout,
+        stderr=_TIMEOUT_TEMPLATE.stderr,
+    )
+
+
+def test_v2_db_push_retries_a_timeout(monkeypatch, tmp_path):
+    """v2: a `prisma db push` that times out is retried, not turned into a
+    TypeError by classifying its bytes stderr as if it were text."""
+    _prepare_v2_resolver(monkeypatch, tmp_path)
+
+    def _side_effect(n, cmd):
+        if n == 1:
+            raise _real_timeout_expired(cmd)
+        return _DeployApplied()
+
+    monkeypatch.setattr(
+        ProxyExtrasDBManager, "spend_logs_is_partitioned", lambda: False
+    )
+    run, pushes = _db_push_only(_side_effect)
+    with patch("subprocess.run", side_effect=run):
+        ok = ProxyExtrasDBManager.setup_database(use_migrate=False, use_v2_resolver=True)
+
+    assert ok is True
+    assert pushes["n"] == 2
+
+
+def test_v2_db_push_timeouts_are_bounded(monkeypatch, tmp_path):
+    """v2: a `prisma db push` that never stops timing out gives up as a
+    RuntimeError, which is the only exception proxy_cli.py exits cleanly on."""
+    _prepare_v2_resolver(monkeypatch, tmp_path)
+
+    def _side_effect(n, cmd):
+        raise _real_timeout_expired(cmd)
+
+    monkeypatch.setattr(
+        ProxyExtrasDBManager, "spend_logs_is_partitioned", lambda: False
+    )
+    run, pushes = _db_push_only(_side_effect)
+    with patch("subprocess.run", side_effect=run):
+        with pytest.raises(RuntimeError, match=r"prisma db push failed after \d+"):
+            ProxyExtrasDBManager.setup_database(use_migrate=False, use_v2_resolver=True)
+
+    assert pushes["n"] == _PRISMA_ATTEMPTS
+
+
+def test_v2_unclassified_failure_is_not_treated_as_transient(monkeypatch, tmp_path):
+    """v2: an unrecognised deploy failure still raises on the first attempt."""
+    _prepare_v2_resolver(monkeypatch, tmp_path)
+
+    def _side_effect(n, cmd):
+        raise subprocess.CalledProcessError(
+            returncode=1,
+            cmd=cmd,
+            stderr="Error: relation \"LiteLLM_SpendLogs\" does not exist",
+            output="",
+        )
+
+    run, deploys = _deploy_only(_side_effect)
+    with patch("subprocess.run", side_effect=run):
+        with pytest.raises(RuntimeError, match="cannot be auto-recovered"):
+            ProxyExtrasDBManager.setup_database(use_migrate=True, use_v2_resolver=True)
+
+    assert deploys["n"] == 1
+
+

tests/local_testing/test_basic_python_version.py

@@ -305,14 +305,16 @@ def _run_proxy_server_smoke_test(extra_proxy_args=None):
 
 
 def test_litellm_proxy_server_config_no_general_settings():
-    """Exercises the default (v1) migration resolver."""
+    """Exercises the default (v2) migration resolver."""
     _run_proxy_server_smoke_test()
 
 
-def test_litellm_proxy_server_config_no_general_settings_v2_resolver():
-    """Exercises the opt-in v2 migration resolver.
+def test_litellm_proxy_server_config_no_general_settings_legacy_resolver():
+    """Exercises the legacy (v1) migration resolver against a real database.
 
-    Runs in a separate CI job against a local Postgres to avoid collisions
-    with the v1 variant when they share a database.
+    v2 is the default, so the no-arg test above already covers it. This one is
+    the only place the v1 opt-out gets real-DB migration plus proxy-boot
+    coverage, and it runs in a separate CI job against its own Postgres to
+    avoid collisions with the default variant.
     """
-    _run_proxy_server_smoke_test(extra_proxy_args=["--use_v2_migration_resolver"])
+    _run_proxy_server_smoke_test(extra_proxy_args=["--use_legacy_migration_resolver"])

tests/pass_through_unit_tests/test_websearch_interception_e2e.py

@@ -937,6 +937,14 @@ async def test_pre_request_hook_modifies_request_body():
 
     print("✅ WebSearchInterceptionLogger initialized")
 
+    mock_router = MagicMock()
+    mock_router.search_tools = [
+        {
+            "search_tool_name": "test-search-tool",
+            "litellm_params": {"search_provider": "tavily"},
+        }
+    ]
+
     # Track what actually gets sent to the API
     captured_request = {}
 
@@ -987,6 +995,9 @@ async def test_pre_request_hook_modifies_request_body():
     with patch(
         "litellm.llms.anthropic.experimental_pass_through.messages.handler.anthropic_messages_handler",
         side_effect=mock_anthropic_messages_handler,
+    ), patch(  # test-quality-ok: the hook imports this process-global router at call time; no injection seam exists to register search_tools
+        "litellm.proxy.proxy_server.llm_router",
+        mock_router,
     ):
 
         print(

tests/test_litellm/integrations/websearch_interception/test_websearch_interception_handler.py

@@ -14,7 +14,7 @@ from litellm.integrations.websearch_interception.handler import (
 )
 from litellm.llms.base_llm.search.transformation import SearchResponse
 from litellm.proxy._types import LiteLLM_ObjectPermissionTable, LiteLLM_TeamTable, ProxyException, UserAPIKeyAuth
-from litellm.types.utils import CallTypes, LlmProviders
+from litellm.types.utils import LlmProviders
 
 
 def test_initialize_from_proxy_config():
@@ -230,124 +230,6 @@ async def test_execute_search_passes_selected_search_tool_litellm_params(monkeyp
     assert forwarded_kwargs["max_retries"] == 2
 
 
-@pytest.mark.asyncio
-@pytest.mark.parametrize(
-    ("search_tools", "error"),
-    [
-        pytest.param(None, "was not found", id="router-not-configured"),
-        pytest.param(
-            [{"search_tool_name": "other-search", "litellm_params": {"search_provider": "tavily"}}],
-            "was not found",
-            id="requested-tool-not-configured",
-        ),
-        pytest.param(
-            [{"search_tool_name": "parallel-search", "litellm_params": "not-a-mapping"}],
-            "does not define a valid search provider",
-            id="invalid-parameters",
-        ),
-        pytest.param(
-            [{"search_tool_name": "parallel-search", "litellm_params": {}}],
-            "does not define a valid search provider",
-            id="missing-provider",
-        ),
-        pytest.param(
-            [{"search_tool_name": "parallel-search", "litellm_params": {"search_provider": "   "}}],
-            "does not define a valid search provider",
-            id="whitespace-provider",
-        ),
-        pytest.param(
-            [{"search_tool_name": "parallel-search", "litellm_params": {"search_provider": 123}}],
-            "does not define a valid search provider",
-            id="invalid-provider",
-        ),
-    ],
-)
-async def test_execute_search_rejects_invalid_explicit_search_tool(monkeypatch, search_tools, error):
-    import litellm
-    from litellm.proxy import proxy_server
-
-    logger = WebSearchInterceptionLogger(search_tool_name="parallel-search")
-    router = None if search_tools is None else MagicMock(search_tools=search_tools)
-    mock_asearch = AsyncMock()
-
-    monkeypatch.setattr(proxy_server, "llm_router", router)
-    monkeypatch.setattr(litellm, "asearch", mock_asearch)
-
-    with pytest.raises(ValueError, match=f"Configured search tool 'parallel-search' {error}"):
-        await logger._execute_search("what is litellm")
-
-    mock_asearch.assert_not_awaited()
-
-
-@pytest.mark.asyncio
-async def test_execute_search_honors_explicit_parallel_search_tool(monkeypatch):
-    import litellm
-    from litellm.proxy import proxy_server
-
-    logger = WebSearchInterceptionLogger(search_tool_name="parallel-search")
-    router = MagicMock(
-        search_tools=[
-            {
-                "search_tool_name": "other-search",
-                "litellm_params": {"search_provider": "tavily", "api_key": "other-key"},
-            },
-            {
-                "search_tool_name": "parallel-search",
-                "litellm_params": {"search_provider": "parallel_ai", "api_key": "parallel-key"},
-            },
-        ],
-    )
-    mock_asearch = AsyncMock(return_value=SearchResponse(object="search", results=[]))
-
-    monkeypatch.setattr(proxy_server, "llm_router", router)
-    monkeypatch.setattr(litellm, "asearch", mock_asearch)
-
-    await logger._execute_search("what is litellm")
-
-    mock_asearch.assert_awaited_once_with(
-        query="what is litellm",
-        search_provider="parallel_ai",
-        api_key="parallel-key",
-    )
-
-
-@pytest.mark.asyncio
-@pytest.mark.parametrize(
-    ("search_tools", "expected_search_kwargs"),
-    [
-        pytest.param(None, {"search_provider": "perplexity"}, id="router-not-configured"),
-        pytest.param(
-            [
-                {
-                    "search_tool_name": "first-search",
-                    "litellm_params": {"search_provider": "tavily", "api_key": "first-key"},
-                },
-                {
-                    "search_tool_name": "parallel-search",
-                    "litellm_params": {"search_provider": "parallel_ai", "api_key": "parallel-key"},
-                },
-            ],
-            {"search_provider": "tavily", "api_key": "first-key"},
-            id="first-configured-tool",
-        ),
-    ],
-)
-async def test_execute_search_preserves_implicit_provider_selection(monkeypatch, search_tools, expected_search_kwargs):
-    import litellm
-    from litellm.proxy import proxy_server
-
-    logger = WebSearchInterceptionLogger()
-    router = None if search_tools is None else MagicMock(search_tools=search_tools)
-    mock_asearch = AsyncMock(return_value=SearchResponse(object="search", results=[]))
-
-    monkeypatch.setattr(proxy_server, "llm_router", router)
-    monkeypatch.setattr(litellm, "asearch", mock_asearch)
-
-    await logger._execute_search("what is litellm")
-
-    mock_asearch.assert_awaited_once_with(query="what is litellm", **expected_search_kwargs)
-
-
 @pytest.mark.asyncio
 async def test_execute_search_attributes_spend_to_the_calling_key(monkeypatch):
     """An intercepted search is billed and logged against the key that made the LLM request.
@@ -515,72 +397,6 @@ async def test_execute_search_enforces_team_search_tool_permission(monkeypatch):
     mock_asearch.assert_not_awaited()
 
 
-@pytest.mark.asyncio
-@pytest.mark.parametrize(
-    ("call_type", "web_search_tool"),
-    [
-        pytest.param(
-            CallTypes.acompletion,
-            {"type": "web_search_20250305", "name": "web_search"},
-            id="chat-completion",
-        ),
-        pytest.param(CallTypes.responses, {"type": "web_search"}, id="responses"),
-        pytest.param(CallTypes.aresponses, {"type": "web_search"}, id="async-responses"),
-        pytest.param(
-            CallTypes.anthropic_messages,
-            {"type": "web_search_20250305", "name": "web_search"},
-            id="anthropic-messages",
-        ),
-    ],
-)
-async def test_deployment_hook_dispatcher_propagates_missing_explicit_search_tool(
-    monkeypatch, call_type, web_search_tool
-):
-    import litellm
-    from litellm.proxy import proxy_server
-    from litellm.utils import async_pre_call_deployment_hook
-
-    logger = WebSearchInterceptionLogger(enabled_providers=["bedrock"], search_tool_name="parallel-search")
-    mock_asearch = AsyncMock()
-    kwargs = {
-        "model": "bedrock/claude-sonnet-4",
-        "tools": [web_search_tool],
-        "custom_llm_provider": "bedrock",
-    }
-
-    monkeypatch.setattr(
-        proxy_server,
-        "llm_router",
-        MagicMock(search_tools=[{"search_tool_name": "other-search", "litellm_params": {"search_provider": "tavily"}}]),
-    )
-    monkeypatch.setattr(litellm, "callbacks", [logger])
-    monkeypatch.setattr(litellm, "asearch", mock_asearch)
-
-    with pytest.raises(ValueError, match="Configured search tool 'parallel-search' was not found"):
-        await async_pre_call_deployment_hook(kwargs=kwargs, call_type=call_type.value)
-
-    assert kwargs["tools"] == [web_search_tool]
-    mock_asearch.assert_not_awaited()
-
-
-@pytest.mark.asyncio
-async def test_deployment_hook_skips_explicit_tool_validation_for_non_search_responses(monkeypatch):
-    from litellm.proxy import proxy_server
-
-    logger = WebSearchInterceptionLogger(enabled_providers=["bedrock"], search_tool_name="parallel-search")
-    monkeypatch.setattr(proxy_server, "llm_router", MagicMock(search_tools=[]))
-
-    result = await logger.async_pre_call_deployment_hook(
-        kwargs={
-            "tools": [{"type": "function", "name": "calculator"}],
-            "custom_llm_provider": "bedrock",
-        },
-        call_type=CallTypes.aresponses,
-    )
-
-    assert result is None
-
-
 @pytest.mark.asyncio
 async def test_async_pre_call_deployment_hook_provider_from_top_level_kwargs():
     """Test that async_pre_call_deployment_hook finds custom_llm_provider at top-level kwargs.

tests/test_litellm/litellm_core_utils/test_streaming_handler.py

@@ -4692,3 +4692,82 @@ async def test_async_stream_assembled_response_keeps_vertex_traffic_type(logging
     assembled = litellm.stream_chunk_builder(chunks=received, messages=[{"role": "user", "content": "hi"}])
     assert assembled is not None
     assert assembled._hidden_params["provider_specific_fields"]["traffic_type"] == "ON_DEMAND_FLEX"
+
+
+class TestStableStreamingResponseId:
+    """
+    All chunks of one streamed response must share the same top-level id
+    (OpenAI streaming contract). Providers streaming via GenericStreamingChunk
+    (e.g. GigaChat) do not propagate an upstream response id, so
+    CustomStreamWrapper must pin the id from the first chunk it creates,
+    mirroring the existing `created` pinning (issue #11437).
+
+    Clients such as goose merge streamed deltas into one assistant message by
+    chunk id; per-chunk ids split a single reply into many messages.
+    """
+
+    def test_generic_chunks_share_one_id(self):
+        def _generic_chunks():
+            return iter(
+                [
+                    {
+                        "text": "Hello",
+                        "tool_use": None,
+                        "is_finished": False,
+                        "finish_reason": "",
+                        "usage": None,
+                        "index": 0,
+                    },
+                    {
+                        "text": " world",
+                        "tool_use": None,
+                        "is_finished": False,
+                        "finish_reason": "",
+                        "usage": None,
+                        "index": 0,
+                    },
+                    {
+                        "text": "",
+                        "tool_use": None,
+                        "is_finished": True,
+                        "finish_reason": "stop",
+                        "usage": {
+                            "prompt_tokens": 1,
+                            "completion_tokens": 2,
+                            "total_tokens": 3,
+                        },
+                        "index": 0,
+                    },
+                ]
+            )
+
+        wrapper = CustomStreamWrapper(
+            completion_stream=_generic_chunks(),
+            model="gigachat/GigaChat-2-Max",
+            logging_obj=MagicMock(),
+            custom_llm_provider="gigachat",
+        )
+        ids = [chunk.id for chunk in wrapper if chunk.id]
+        assert ids, "no chunks emitted"
+        assert len(set(ids)) == 1, f"chunk ids differ across one stream: {ids}"
+
+    def test_creator_pins_id_from_first_chunk(self):
+        wrapper = CustomStreamWrapper(
+            completion_stream=iter([]),
+            model="gigachat/GigaChat-2-Max",
+            logging_obj=MagicMock(),
+            custom_llm_provider="gigachat",
+        )
+        first = wrapper.model_response_creator()
+        assert wrapper.response_id == first.id
+        assert wrapper.model_response_creator().id == first.id
+
+    def test_provider_supplied_id_still_wins(self):
+        wrapper = CustomStreamWrapper(
+            completion_stream=iter([]),
+            model="gigachat/GigaChat-2-Max",
+            logging_obj=MagicMock(),
+            custom_llm_provider="gigachat",
+        )
+        wrapper.response_id = "chatcmpl-from-provider"
+        assert wrapper.model_response_creator().id == "chatcmpl-from-provider"

tests/test_litellm/llms/anthropic/chat/test_anthropic_chat_transformation.py

@@ -6312,3 +6312,45 @@ def test_forced_tool_choice_gating_driven_by_model_map_flag(local_model_cost_map
     )
 
     assert result["tool_choice"] == {"type": "auto"}
+
+
+def test_anthropic_drop_params_keeps_format_only_output_config(monkeypatch):
+    """``drop_params=True`` must not consume ``output_config.format``: the drop
+    gate is an effort gate and ``format`` is a structured-output field."""
+    monkeypatch.setattr(litellm, "drop_params", True)
+    config = AnthropicConfig()
+    schema_format = {
+        "type": "json_schema",
+        "schema": {"type": "object", "properties": {"z": {"type": "integer"}}},
+    }
+
+    result = config.transform_request(
+        model="claude-3-haiku-20240307",
+        messages=[{"role": "user", "content": "Hello"}],
+        optional_params={"output_config": {"format": schema_format}},
+        litellm_params={},
+        headers={},
+    )
+
+    assert result.get("output_config") == {"format": schema_format}
+
+
+def test_anthropic_drop_params_reduces_mixed_output_config_to_format(monkeypatch):
+    """``drop_params=True`` drops the effort key on unsupported models but keeps
+    ``format`` so structured outputs still reach the provider."""
+    monkeypatch.setattr(litellm, "drop_params", True)
+    config = AnthropicConfig()
+    schema_format = {
+        "type": "json_schema",
+        "schema": {"type": "object", "properties": {"z": {"type": "integer"}}},
+    }
+
+    result = config.transform_request(
+        model="claude-3-haiku-20240307",
+        messages=[{"role": "user", "content": "Hello"}],
+        optional_params={"output_config": {"effort": "low", "format": schema_format}},
+        litellm_params={},
+        headers={},
+    )
+
+    assert result.get("output_config") == {"format": schema_format}

tests/test_litellm/llms/bedrock/chat/invoke_transformations/test_bedrock_chat_invoke_transformations_anthropic_claude3_transformation.py

@@ -428,30 +428,58 @@ def test_output_config_forwarded_for_bedrock_chat_invoke_request():
 
 
 def test_output_config_format_converted_for_bedrock_chat_invoke_request():
-    """Bedrock Invoke chat path consumes ``output_config.format`` before forwarding."""
+    """Bedrock Invoke chat path inlines ``output_config.format`` for models
+    without native structured-output support and keeps the effort key."""
     config = AmazonAnthropicClaudeConfig()
     schema = {
         "type": "object",
         "properties": {"answer": {"type": "string"}},
     }
 
-    result = config.transform_request(
+    with patch(  # test-quality-ok: pin non-native path
+        "litellm.llms.bedrock.common_utils._bedrock_model_supports",
+        side_effect=lambda _model, key: key == "supports_output_config",
+    ):
+        result = config.transform_request(
+            model="anthropic.claude-opus-4-7",
+            messages=[{"role": "user", "content": "test"}],
+            optional_params={
+                "max_tokens": 100,
+                "output_config": {
+                    "effort": "xhigh",
+                    "format": {"type": "json_schema", "schema": schema},
+                },
+            },
+            litellm_params={},
+            headers={},
+        )
+
+    assert result.get("output_config") == {"effort": "xhigh"}
+    last_content = result["messages"][0]["content"]
+    assert json.loads(last_content[-1]["text"]) == schema
+
+
+def test_output_config_format_forwarded_for_bedrock_chat_invoke_request():
+    """Bedrock Invoke chat path forwards ``output_config.format`` alongside effort
+    for models with native structured-output support (Claude Opus 4.7)."""
+    schema_format = {
+        "type": "json_schema",
+        "schema": {"type": "object", "properties": {"answer": {"type": "string"}}},
+    }
+
+    result = AmazonAnthropicClaudeConfig().transform_request(
         model="anthropic.claude-opus-4-7",
         messages=[{"role": "user", "content": "test"}],
         optional_params={
             "max_tokens": 100,
-            "output_config": {
-                "effort": "xhigh",
-                "format": {"type": "json_schema", "schema": schema},
-            },
+            "output_config": {"effort": "xhigh", "format": schema_format},
         },
         litellm_params={},
         headers={},
     )
 
-    assert result.get("output_config") == {"effort": "xhigh"}
-    last_content = result["messages"][0]["content"]
-    assert json.loads(last_content[-1]["text"]) == schema
+    assert result.get("output_config") == {"effort": "xhigh", "format": schema_format}
+    assert "answer" not in json.dumps(result["messages"])
 
 
 @pytest.mark.parametrize(
@@ -488,7 +516,7 @@ def test_bedrock_chat_invoke_checks_output_config_support_with_bedrock_provider(
     optional_params = {"max_tokens": 100, "output_config": {"effort": "high"}}
 
     with patch(
-        "litellm.llms.bedrock.chat.invoke_transformations.anthropic_claude3_transformation._supports_factory",
+        "litellm.llms.bedrock.common_utils._bedrock_model_supports",
         return_value=True,
     ) as mock_supports_factory:
         result = config.transform_request(
@@ -499,11 +527,7 @@ def test_bedrock_chat_invoke_checks_output_config_support_with_bedrock_provider(
             headers={},
         )
 
-    mock_supports_factory.assert_called_once_with(
-        model="us.anthropic.claude-opus-4-7",
-        custom_llm_provider="bedrock",
-        key="supports_output_config",
-    )
+    mock_supports_factory.assert_called_once_with("us.anthropic.claude-opus-4-7", "supports_output_config")
     assert result["output_config"] == {"effort": "high"}
 
 
@@ -542,3 +566,80 @@ def test_output_format_removed_from_bedrock_invoke_request():
     assert (
         "output_format" not in result
     ), f"output_format should be removed for Bedrock Invoke, got keys: {result.keys()}"
+
+
+def test_bedrock_chat_invoke_forwards_output_config_format_natively(local_model_cost_map):
+    """Regression: ``output_config.format`` is forwarded verbatim on models Bedrock
+    enforces structured outputs for, instead of being inlined as prompt text."""
+    import json
+
+    config = AmazonAnthropicClaudeConfig()
+    schema_format = {
+        "type": "json_schema",
+        "schema": {
+            "type": "object",
+            "properties": {"zebra_count": {"type": "integer"}},
+            "required": ["zebra_count"],
+            "additionalProperties": False,
+        },
+    }
+
+    result = config.transform_request(
+        model="us.anthropic.claude-sonnet-4-5-20250929-v1:0",
+        messages=[{"role": "user", "content": "say hello"}],
+        optional_params={
+            "max_tokens": 100,
+            "output_config": {"format": schema_format},
+        },
+        litellm_params={},
+        headers={},
+    )
+
+    assert result.get("output_config") == {"format": schema_format}
+    assert "zebra_count" not in json.dumps(result["messages"])
+
+
+def test_bedrock_chat_invoke_drop_params_keeps_native_output_config_format(local_model_cost_map, monkeypatch):
+    """``drop_params=True`` must not eat ``output_config.format`` before the
+    native-forwarding router runs (Sonnet 4.5 has no effort flags)."""
+    import litellm
+
+    monkeypatch.setattr(litellm, "drop_params", True)
+    schema_format = {
+        "type": "json_schema",
+        "schema": {"type": "object", "properties": {"zebra_count": {"type": "integer"}}},
+    }
+
+    result = AmazonAnthropicClaudeConfig().transform_request(
+        model="us.anthropic.claude-sonnet-4-5-20250929-v1:0",
+        messages=[{"role": "user", "content": "say hello"}],
+        optional_params={"max_tokens": 100, "output_config": {"format": schema_format}},
+        litellm_params={},
+        headers={},
+    )
+
+    assert result.get("output_config") == {"format": schema_format}
+
+
+def test_bedrock_chat_invoke_drop_params_still_inlines_for_non_native(local_model_cost_map, monkeypatch):
+    """``drop_params=True`` on a model without native structured-output support
+    still reaches the inline-schema fallback instead of losing the schema."""
+    import litellm
+
+    monkeypatch.setattr(litellm, "drop_params", True)
+    schema = {"type": "object", "properties": {"zebra_count": {"type": "integer"}}}
+
+    result = AmazonAnthropicClaudeConfig().transform_request(
+        model="anthropic.claude-3-haiku-20240307-v1:0",
+        messages=[{"role": "user", "content": "say hello"}],
+        optional_params={
+            "max_tokens": 100,
+            "output_config": {"format": {"type": "json_schema", "schema": schema}},
+        },
+        litellm_params={},
+        headers={},
+    )
+
+    assert "output_config" not in result
+    last_content = result["messages"][-1]["content"]
+    assert json.loads(last_content[-1]["text"]) == schema

tests/test_litellm/llms/bedrock/messages/invoke_transformations/test_anthropic_claude3_transformation.py

@@ -935,7 +935,7 @@ def test_bedrock_messages_strips_output_config():
     }
 
     with patch(
-        "litellm.llms.bedrock.messages.invoke_transformations.anthropic_claude3_transformation._supports_factory",
+        "litellm.llms.bedrock.common_utils._bedrock_model_supports",
         return_value=False,
     ):
         result = cfg.transform_anthropic_messages_request(
@@ -970,7 +970,7 @@ def test_bedrock_messages_preserves_output_config_for_claude_4_6():
     }
 
     with patch(
-        "litellm.llms.bedrock.messages.invoke_transformations.anthropic_claude3_transformation._supports_factory",
+        "litellm.llms.bedrock.common_utils._bedrock_model_supports",
         return_value=True,
     ):
         result = cfg.transform_anthropic_messages_request(
@@ -1003,7 +1003,7 @@ def test_bedrock_messages_checks_output_config_support_with_bedrock_provider():
     }
 
     with patch(
-        "litellm.llms.bedrock.messages.invoke_transformations.anthropic_claude3_transformation._supports_factory",
+        "litellm.llms.bedrock.common_utils._bedrock_model_supports",
         return_value=True,
     ) as mock_supports_factory:
         result = cfg.transform_anthropic_messages_request(
@@ -1014,11 +1014,7 @@ def test_bedrock_messages_checks_output_config_support_with_bedrock_provider():
             headers={},
         )
 
-    mock_supports_factory.assert_called_with(
-        model="us.anthropic.claude-opus-4-7",
-        custom_llm_provider="bedrock",
-        key="supports_output_config",
-    )
+    mock_supports_factory.assert_called_with("us.anthropic.claude-opus-4-7", "supports_output_config")
     assert result["output_config"] == {"effort": "high"}
 
 
@@ -1038,7 +1034,7 @@ def test_bedrock_messages_forwards_output_config():
     }
 
     with patch(
-        "litellm.llms.bedrock.messages.invoke_transformations.anthropic_claude3_transformation._supports_factory",
+        "litellm.llms.bedrock.common_utils._bedrock_model_supports",
         return_value=True,
     ):
         result = cfg.transform_anthropic_messages_request(
@@ -1054,27 +1050,29 @@ def test_bedrock_messages_forwards_output_config():
 
 
 def test_bedrock_messages_forwards_output_config_with_output_format():
-    """``output_config`` is forwarded; ``output_format`` is converted to inline schema."""
+    """Legacy ``output_format`` is forwarded as ``output_config.format`` on models
+    that support native structured outputs, alongside the effort key."""
     from unittest.mock import patch
 
     from litellm.types.router import GenericLiteLLMParams
 
     cfg = AmazonAnthropicClaudeMessagesConfig()
     messages = [{"role": "user", "content": [{"type": "text", "text": "Hello"}]}]
+    schema_format = {
+        "type": "json_schema",
+        "schema": {
+            "type": "object",
+            "properties": {"answer": {"type": "string"}},
+        },
+    }
     optional_params = {
         "max_tokens": 4096,
         "output_config": {"effort": "low"},
-        "output_format": {
-            "type": "json_schema",
-            "schema": {
-                "type": "object",
-                "properties": {"answer": {"type": "string"}},
-            },
-        },
+        "output_format": schema_format,
     }
 
     with patch(
-        "litellm.llms.bedrock.messages.invoke_transformations.anthropic_claude3_transformation._supports_factory",
+        "litellm.llms.bedrock.common_utils._bedrock_model_supports",
         return_value=True,
     ):
         result = cfg.transform_anthropic_messages_request(
@@ -1085,12 +1083,14 @@ def test_bedrock_messages_forwards_output_config_with_output_format():
             headers={},
         )
 
-    assert result.get("output_config") == {"effort": "low"}
+    assert result.get("output_config") == {"effort": "low", "format": schema_format}
     assert "output_format" not in result
+    assert "answer" not in json.dumps(result["messages"])
 
 
 def test_bedrock_messages_converts_output_config_format_to_inline_schema():
-    """``output_config.format`` is consumed so Bedrock does not see an unknown nested key."""
+    """Without native structured-output support, ``output_config.format`` falls back
+    to the inline schema so Bedrock does not see an unknown nested key."""
     from unittest.mock import patch
 
     from litellm.types.router import GenericLiteLLMParams
@@ -1110,8 +1110,8 @@ def test_bedrock_messages_converts_output_config_format_to_inline_schema():
     }
 
     with patch(
-        "litellm.llms.bedrock.messages.invoke_transformations.anthropic_claude3_transformation._supports_factory",
-        return_value=True,
+        "litellm.llms.bedrock.common_utils._bedrock_model_supports",
+        side_effect=lambda _model, key: key == "supports_output_config",
     ):
         result = cfg.transform_anthropic_messages_request(
             model="anthropic.claude-opus-4-7",
@@ -1146,7 +1146,7 @@ def test_bedrock_messages_normalizes_output_config_effort_for_opus(
     cfg = AmazonAnthropicClaudeMessagesConfig()
 
     with patch(
-        "litellm.llms.bedrock.messages.invoke_transformations.anthropic_claude3_transformation._supports_factory",
+        "litellm.llms.bedrock.common_utils._bedrock_model_supports",
         return_value=True,
     ):
         result = cfg.transform_anthropic_messages_request(
@@ -1184,8 +1184,8 @@ def test_bedrock_messages_does_not_mutate_callers_messages_when_embedding_schema
     }
 
     with patch(
-        "litellm.llms.bedrock.messages.invoke_transformations.anthropic_claude3_transformation._supports_factory",
-        return_value=True,
+        "litellm.llms.bedrock.common_utils._bedrock_model_supports",
+        side_effect=lambda _model, key: key == "supports_output_config",
     ):
         result = cfg.transform_anthropic_messages_request(
             model="anthropic.claude-opus-4-7",
@@ -1229,7 +1229,7 @@ def test_bedrock_messages_does_not_mutate_callers_output_config():
     }
 
     with patch(
-        "litellm.llms.bedrock.messages.invoke_transformations.anthropic_claude3_transformation._supports_factory",
+        "litellm.llms.bedrock.common_utils._bedrock_model_supports",
         return_value=True,
     ):
         cfg.transform_anthropic_messages_request(
@@ -1271,7 +1271,7 @@ def test_bedrock_messages_strips_output_config_with_output_format():
     }
 
     with patch(
-        "litellm.llms.bedrock.messages.invoke_transformations.anthropic_claude3_transformation._supports_factory",
+        "litellm.llms.bedrock.common_utils._bedrock_model_supports",
         return_value=False,
     ):
         result = cfg.transform_anthropic_messages_request(
@@ -1332,7 +1332,7 @@ def test_bedrock_messages_drop_params_keeps_output_config_for_4_7():
     litellm.drop_params = True
     try:
         with patch(
-            "litellm.llms.bedrock.messages.invoke_transformations.anthropic_claude3_transformation._supports_factory",
+            "litellm.llms.bedrock.common_utils._bedrock_model_supports",
             return_value=True,
         ):
             result = cfg.transform_anthropic_messages_request(
@@ -1375,7 +1375,7 @@ def test_bedrock_messages_maps_reasoning_effort_for_adaptive_model(
     }
 
     with patch(
-        "litellm.llms.bedrock.messages.invoke_transformations.anthropic_claude3_transformation._supports_factory",
+        "litellm.llms.bedrock.common_utils._bedrock_model_supports",
         return_value=True,
     ):
         result = cfg.transform_anthropic_messages_request(
@@ -1482,7 +1482,7 @@ def test_bedrock_messages_explicit_output_config_wins_over_reasoning_effort():
     }
 
     with patch(
-        "litellm.llms.bedrock.messages.invoke_transformations.anthropic_claude3_transformation._supports_factory",
+        "litellm.llms.bedrock.common_utils._bedrock_model_supports",
         return_value=True,
     ):
         result = cfg.transform_anthropic_messages_request(
@@ -3104,3 +3104,149 @@ async def test_bedrock_sse_wrapper_dispatches_logging_on_client_disconnect():
             break
         await asyncio.sleep(0.01)
     assert logging_obj.completion_start_time is not None
+
+
+def test_bedrock_messages_forwards_output_config_format_natively(local_model_cost_map):
+    """Regression: on a model Bedrock enforces structured outputs for (Claude
+    Sonnet 4.5), ``output_config.format`` must be forwarded verbatim, not
+    silently rewritten into inline prompt text."""
+    from litellm.types.router import GenericLiteLLMParams
+
+    cfg = AmazonAnthropicClaudeMessagesConfig()
+    schema_format = {
+        "type": "json_schema",
+        "schema": {
+            "type": "object",
+            "properties": {
+                "zebra_count": {"type": "integer"},
+                "is_tuesday": {"type": "boolean"},
+            },
+            "required": ["zebra_count", "is_tuesday"],
+            "additionalProperties": False,
+        },
+    }
+
+    result = cfg.transform_anthropic_messages_request(
+        model="us.anthropic.claude-sonnet-4-5-20250929-v1:0",
+        messages=[{"role": "user", "content": [{"type": "text", "text": "say hello"}]}],
+        anthropic_messages_optional_request_params={
+            "max_tokens": 100,
+            "output_config": {"format": schema_format},
+        },
+        litellm_params=GenericLiteLLMParams(),
+        headers={},
+    )
+
+    assert result.get("output_config") == {"format": schema_format}
+    assert "zebra_count" not in json.dumps(result["messages"])
+
+
+def test_bedrock_messages_inlines_schema_for_claude_5(local_model_cost_map):
+    """Bedrock rejects ``output_config.format`` for the Claude 5 family, so the
+    schema falls back to the inline-text path instead of a deterministic 400."""
+    from litellm.types.router import GenericLiteLLMParams
+
+    cfg = AmazonAnthropicClaudeMessagesConfig()
+    schema = {
+        "type": "object",
+        "properties": {"zebra_count": {"type": "integer"}},
+    }
+
+    result = cfg.transform_anthropic_messages_request(
+        model="us.anthropic.claude-sonnet-5",
+        messages=[{"role": "user", "content": [{"type": "text", "text": "say hello"}]}],
+        anthropic_messages_optional_request_params={
+            "max_tokens": 100,
+            "output_config": {"format": {"type": "json_schema", "schema": schema}},
+        },
+        litellm_params=GenericLiteLLMParams(),
+        headers={},
+    )
+
+    assert "output_config" not in result
+    last_content = result["messages"][-1]["content"]
+    assert json.loads(last_content[-1]["text"]) == schema
+
+
+def test_bedrock_messages_legacy_output_format_wins_over_output_config_format(local_model_cost_map):
+    """When a request carries both schema forms, the legacy top-level
+    ``output_format`` keeps winning, matching the pre-existing precedence."""
+    from litellm.types.router import GenericLiteLLMParams
+
+    cfg = AmazonAnthropicClaudeMessagesConfig()
+    legacy_format = {
+        "type": "json_schema",
+        "schema": {"type": "object", "properties": {"legacy_field": {"type": "string"}}},
+    }
+    newer_format = {
+        "type": "json_schema",
+        "schema": {"type": "object", "properties": {"newer_field": {"type": "string"}}},
+    }
+
+    result = cfg.transform_anthropic_messages_request(
+        model="us.anthropic.claude-sonnet-4-5-20250929-v1:0",
+        messages=[{"role": "user", "content": [{"type": "text", "text": "say hello"}]}],
+        anthropic_messages_optional_request_params={
+            "max_tokens": 100,
+            "output_format": legacy_format,
+            "output_config": {"format": newer_format},
+        },
+        litellm_params=GenericLiteLLMParams(),
+        headers={},
+    )
+
+    assert result.get("output_config") == {"format": legacy_format}
+    assert "output_format" not in result
+    assert "newer_field" not in json.dumps(result)
+
+
+def test_bedrock_messages_drop_params_keeps_native_output_config_format(local_model_cost_map, monkeypatch):
+    """``drop_params=True`` must not strip a natively forwarded
+    ``output_config.format`` on models without effort support (Sonnet 4.5)."""
+    import litellm
+    from litellm.types.router import GenericLiteLLMParams
+
+    monkeypatch.setattr(litellm, "drop_params", True)
+    cfg = AmazonAnthropicClaudeMessagesConfig()
+    schema_format = {
+        "type": "json_schema",
+        "schema": {"type": "object", "properties": {"zebra_count": {"type": "integer"}}},
+    }
+
+    result = cfg.transform_anthropic_messages_request(
+        model="us.anthropic.claude-sonnet-4-5-20250929-v1:0",
+        messages=[{"role": "user", "content": [{"type": "text", "text": "say hello"}]}],
+        anthropic_messages_optional_request_params={
+            "max_tokens": 100,
+            "output_config": {"format": schema_format},
+        },
+        litellm_params=GenericLiteLLMParams(),
+        headers={},
+    )
+
+    assert result.get("output_config") == {"format": schema_format}
+
+
+def test_bedrock_messages_strips_effort_but_keeps_format_for_sonnet_4_5(local_model_cost_map):
+    """Sonnet 4.5 has native structured-output support but no effort support, so
+    a mixed ``output_config`` keeps ``format`` and drops ``effort``."""
+    from litellm.types.router import GenericLiteLLMParams
+
+    cfg = AmazonAnthropicClaudeMessagesConfig()
+    schema_format = {
+        "type": "json_schema",
+        "schema": {"type": "object", "properties": {"zebra_count": {"type": "integer"}}},
+    }
+
+    result = cfg.transform_anthropic_messages_request(
+        model="us.anthropic.claude-sonnet-4-5-20250929-v1:0",
+        messages=[{"role": "user", "content": [{"type": "text", "text": "say hello"}]}],
+        anthropic_messages_optional_request_params={
+            "max_tokens": 4096,
+            "output_config": {"format": schema_format, "effort": "high"},
+        },
+        litellm_params=GenericLiteLLMParams(),
+        headers={},
+    )
+
+    assert result.get("output_config") == {"format": schema_format}

tests/test_litellm/llms/bedrock/test_bedrock_common_utils.py

@@ -522,6 +522,47 @@ def test_merge_bedrock_aws_request_params_keeps_caller_credentials_without_stati
     assert merged["aws_region_name"] == "us-west-2"
 
 
+def test_strip_unsupported_output_config_keeps_format_drops_effort(local_model_cost_map):
+    """On a model with neither effort flag, only the ``format`` key survives."""
+    from litellm.llms.bedrock.common_utils import (
+        strip_unsupported_bedrock_invoke_output_config_keys,
+    )
+
+    schema_format = {"type": "json_schema", "schema": {"type": "object"}}
+    body = {"output_config": {"effort": "high", "format": schema_format}}
+
+    strip_unsupported_bedrock_invoke_output_config_keys(
+        model="anthropic.claude-3-haiku-20240307-v1:0",
+        request_body=body,
+    )
+
+    assert body["output_config"] == {"format": schema_format}
+
+
+def test_apply_structured_output_prefers_legacy_output_format(local_model_cost_map):
+    """The legacy ``output_format`` wins over ``output_config.format`` when a
+    request carries both, matching the pre-existing precedence."""
+    from litellm.llms.bedrock.common_utils import (
+        apply_bedrock_invoke_structured_output,
+    )
+
+    legacy = {"type": "json_schema", "schema": {"type": "object", "properties": {"a": {"type": "string"}}}}
+    newer = {"type": "json_schema", "schema": {"type": "object", "properties": {"b": {"type": "string"}}}}
+    body = {
+        "messages": [{"role": "user", "content": "hi"}],
+        "output_format": legacy,
+        "output_config": {"format": newer},
+    }
+
+    apply_bedrock_invoke_structured_output(
+        model="us.anthropic.claude-sonnet-4-5-20250929-v1:0",
+        request_body=body,
+    )
+
+    assert body["output_config"] == {"format": legacy}
+    assert "output_format" not in body
+
+
 def test_sign_aws_request_assumes_role_with_external_id(monkeypatch):
     """A trust policy requiring sts:ExternalId must be satisfied when signing batch API requests."""
     import datetime

tests/test_litellm/llms/dashscope/test_qwen_brand_aliases.py

@@ -0,0 +1,331 @@
+import math
+
+import pytest
+
+import litellm
+from litellm import completion, get_llm_provider
+from litellm.llms.dashscope.chat.transformation import DashScopeChatConfig
+from litellm.llms.dashscope.cost_calculator import (
+    cost_per_token as dashscope_cost_per_token,
+)
+from litellm.llms.dashscope.embed.transformation import DashScopeEmbeddingConfig
+from litellm.llms.dashscope.image_generation.transformation import (
+    DashScopeImageGenerationConfig,
+)
+from litellm.llms.dashscope.qwen_ai_platform import (
+    QWEN_AI_PLATFORM_API_BASE,
+    QWEN_AI_PLATFORM_IMAGE_API_BASE,
+    QWEN_AI_PLATFORM_RERANK_API_BASE,
+    QwenAIPlatformChatConfig,
+    QwenAIPlatformEmbeddingConfig,
+    QwenAIPlatformImageGenerationConfig,
+    QwenAIPlatformRerankConfig,
+)
+from litellm.llms.dashscope.qwencloud import (
+    QWENCLOUD_API_BASE,
+    QWENCLOUD_IMAGE_API_BASE,
+    QWENCLOUD_RERANK_API_BASE,
+    QwenCloudChatConfig,
+    QwenCloudEmbeddingConfig,
+    QwenCloudImageGenerationConfig,
+    QwenCloudRerankConfig,
+)
+from litellm.llms.dashscope.rerank.transformation import DashScopeRerankConfig
+from litellm.types.utils import LlmProviders, Usage
+from litellm.utils import ProviderConfigManager
+
+DASHSCOPE_FAMILY_ENV_VARS = [
+    "DASHSCOPE_API_KEY",
+    "DASHSCOPE_API_BASE",
+    "DASHSCOPE_API_BASE_RERANK",
+    "DASHSCOPE_API_BASE_IMAGE",
+    "QWENCLOUD_API_KEY",
+    "QWENCLOUD_API_BASE",
+    "QWENCLOUD_API_BASE_RERANK",
+    "QWENCLOUD_API_BASE_IMAGE",
+    "QWEN_AI_PLATFORM_API_KEY",
+    "QWEN_AI_PLATFORM_API_BASE",
+    "QWEN_AI_PLATFORM_API_BASE_RERANK",
+    "QWEN_AI_PLATFORM_API_BASE_IMAGE",
+]
+
+BRAND_CASES = [
+    pytest.param(
+        {
+            "provider": "qwencloud",
+            "enum": LlmProviders.QWENCLOUD,
+            "key_env": "QWENCLOUD_API_KEY",
+            "base_env": "QWENCLOUD_API_BASE",
+            "default_base": QWENCLOUD_API_BASE,
+            "default_rerank_base": QWENCLOUD_RERANK_API_BASE,
+            "default_image_base": QWENCLOUD_IMAGE_API_BASE,
+            "chat_config": QwenCloudChatConfig,
+            "embedding_config": QwenCloudEmbeddingConfig,
+            "rerank_config": QwenCloudRerankConfig,
+            "image_config": QwenCloudImageGenerationConfig,
+        },
+        id="qwencloud",
+    ),
+    pytest.param(
+        {
+            "provider": "qwen_ai_platform",
+            "enum": LlmProviders.QWEN_AI_PLATFORM,
+            "key_env": "QWEN_AI_PLATFORM_API_KEY",
+            "base_env": "QWEN_AI_PLATFORM_API_BASE",
+            "default_base": QWEN_AI_PLATFORM_API_BASE,
+            "default_rerank_base": QWEN_AI_PLATFORM_RERANK_API_BASE,
+            "default_image_base": QWEN_AI_PLATFORM_IMAGE_API_BASE,
+            "chat_config": QwenAIPlatformChatConfig,
+            "embedding_config": QwenAIPlatformEmbeddingConfig,
+            "rerank_config": QwenAIPlatformRerankConfig,
+            "image_config": QwenAIPlatformImageGenerationConfig,
+        },
+        id="qwen_ai_platform",
+    ),
+]
+
+
+@pytest.fixture(autouse=True)
+def clear_dashscope_family_env(monkeypatch):
+    for env_var in DASHSCOPE_FAMILY_ENV_VARS:
+        monkeypatch.delenv(env_var, raising=False)
+
+
+class TestQwenBrandProviderResolution:
+    @pytest.mark.parametrize("brand", BRAND_CASES)
+    def test_get_llm_provider_resolves_brand_default_base(self, brand):
+        model, provider, api_key, api_base = get_llm_provider(f"{brand['provider']}/qwen-max", api_key="sk-explicit")
+        assert model == "qwen-max"
+        assert provider == brand["provider"]
+        assert api_key == "sk-explicit"
+        assert api_base == brand["default_base"]
+
+    def test_dashscope_resolution_unchanged(self):
+        model, provider, api_key, api_base = get_llm_provider("dashscope/qwen-max", api_key="sk-explicit")
+        assert model == "qwen-max"
+        assert provider == "dashscope"
+        assert api_base == "https://dashscope.aliyuncs.com/compatible-mode/v1"
+
+    @pytest.mark.parametrize("brand", BRAND_CASES)
+    def test_brand_env_key_wins_over_dashscope_key(self, monkeypatch, brand):
+        monkeypatch.setenv(brand["key_env"], "sk-brand")
+        monkeypatch.setenv("DASHSCOPE_API_KEY", "sk-dashscope")
+        _, _, api_key, _ = get_llm_provider(f"{brand['provider']}/qwen-max")
+        assert api_key == "sk-brand"
+
+    @pytest.mark.parametrize("brand", BRAND_CASES)
+    def test_dashscope_key_is_fallback(self, monkeypatch, brand):
+        monkeypatch.setenv("DASHSCOPE_API_KEY", "sk-dashscope")
+        _, _, api_key, _ = get_llm_provider(f"{brand['provider']}/qwen-max")
+        assert api_key == "sk-dashscope"
+
+    @pytest.mark.parametrize("brand", BRAND_CASES)
+    def test_dashscope_api_base_does_not_leak_into_brand(self, monkeypatch, brand):
+        monkeypatch.setenv("DASHSCOPE_API_BASE", "https://legacy.example.com/v1")
+        _, _, _, api_base = get_llm_provider(f"{brand['provider']}/qwen-max", api_key="sk-explicit")
+        assert api_base == brand["default_base"]
+
+    @pytest.mark.parametrize("brand", BRAND_CASES)
+    def test_brand_api_base_env_wins(self, monkeypatch, brand):
+        monkeypatch.setenv(brand["base_env"], "https://brand.example.com/v1")
+        _, _, _, api_base = get_llm_provider(f"{brand['provider']}/qwen-max", api_key="sk-explicit")
+        assert api_base == "https://brand.example.com/v1"
+
+
+class TestQwenBrandConfigDispatch:
+    @pytest.mark.parametrize("brand", BRAND_CASES)
+    def test_chat_config(self, brand):
+        config = ProviderConfigManager.get_provider_chat_config("qwen-max", brand["enum"])
+        assert isinstance(config, brand["chat_config"])
+        assert isinstance(config, DashScopeChatConfig)
+
+    @pytest.mark.parametrize("brand", BRAND_CASES)
+    def test_embedding_config(self, brand):
+        config = ProviderConfigManager.get_provider_embedding_config(model="text-embedding-v3", provider=brand["enum"])
+        assert isinstance(config, brand["embedding_config"])
+        assert isinstance(config, DashScopeEmbeddingConfig)
+
+    @pytest.mark.parametrize("brand", BRAND_CASES)
+    def test_rerank_config(self, brand):
+        config = ProviderConfigManager.get_provider_rerank_config(
+            model="gte-rerank-v2",
+            provider=brand["enum"],
+            api_base=None,
+            present_version_params=[],
+        )
+        assert isinstance(config, brand["rerank_config"])
+        assert isinstance(config, DashScopeRerankConfig)
+
+    @pytest.mark.parametrize("brand", BRAND_CASES)
+    def test_image_generation_config(self, brand):
+        config = ProviderConfigManager.get_provider_image_generation_config(model="qwen-image", provider=brand["enum"])
+        assert isinstance(config, brand["image_config"])
+        assert isinstance(config, DashScopeImageGenerationConfig)
+
+
+class TestQwenBrandDefaultUrls:
+    @pytest.mark.parametrize("brand", BRAND_CASES)
+    def test_chat_complete_url(self, brand):
+        url = brand["chat_config"]().get_complete_url(
+            api_base=None,
+            api_key="sk-test",
+            model="qwen-max",
+            optional_params={},
+            litellm_params={},
+        )
+        assert url == f"{brand['default_base']}/chat/completions"
+
+    @pytest.mark.parametrize("brand", BRAND_CASES)
+    def test_embedding_complete_url(self, brand):
+        url = brand["embedding_config"]().get_complete_url(
+            api_base=None,
+            api_key="sk-test",
+            model="text-embedding-v3",
+            optional_params={},
+            litellm_params={},
+        )
+        assert url == f"{brand['default_base']}/embeddings"
+
+    @pytest.mark.parametrize("brand", BRAND_CASES)
+    def test_embedding_ignores_dashscope_api_base(self, monkeypatch, brand):
+        monkeypatch.setenv("DASHSCOPE_API_BASE", "https://legacy.example.com/v1")
+        url = brand["embedding_config"]().get_complete_url(
+            api_base=None,
+            api_key="sk-test",
+            model="text-embedding-v3",
+            optional_params={},
+            litellm_params={},
+        )
+        assert url == f"{brand['default_base']}/embeddings"
+
+    @pytest.mark.parametrize("brand", BRAND_CASES)
+    def test_rerank_complete_url(self, brand):
+        url = brand["rerank_config"]().get_complete_url(api_base=None, model="gte-rerank-v2")
+        assert url == brand["default_rerank_base"]
+
+    @pytest.mark.parametrize("brand", BRAND_CASES)
+    def test_rerank_env_override(self, monkeypatch, brand):
+        monkeypatch.setenv(f"{brand['base_env']}_RERANK", "https://rerank.example.com/v1/reranks")
+        url = brand["rerank_config"]().get_complete_url(api_base=None, model="gte-rerank-v2")
+        assert url == "https://rerank.example.com/v1/reranks"
+
+    @pytest.mark.parametrize("brand", BRAND_CASES)
+    def test_image_generation_complete_url(self, brand):
+        url = brand["image_config"]().get_complete_url(
+            api_base=None,
+            api_key="sk-test",
+            model="qwen-image",
+            optional_params={},
+            litellm_params={},
+        )
+        assert url == brand["default_image_base"]
+
+    @pytest.mark.parametrize("brand", BRAND_CASES)
+    def test_image_generation_ignores_chat_compatible_api_base(self, brand):
+        url = brand["image_config"]().get_complete_url(
+            api_base=brand["default_base"],
+            api_key="sk-test",
+            model="qwen-image",
+            optional_params={},
+            litellm_params={},
+        )
+        assert url == brand["default_image_base"]
+
+    @pytest.mark.parametrize("brand", BRAND_CASES)
+    def test_validate_environment_requires_key(self, brand):
+        with pytest.raises(ValueError, match="DASHSCOPE_API_KEY"):
+            brand["embedding_config"]().validate_environment(
+                headers={},
+                model="text-embedding-v3",
+                messages=[],
+                optional_params={},
+                litellm_params={},
+                api_key=None,
+                api_base=None,
+            )
+
+
+class TestQwenBrandCostParity:
+    @pytest.fixture(autouse=True)
+    def setup_model_cost_map(self, monkeypatch):
+        monkeypatch.setenv("LITELLM_LOCAL_MODEL_COST_MAP", "True")
+        monkeypatch.setattr(litellm, "model_cost", litellm.get_model_cost_map(url=""))
+
+    @pytest.mark.parametrize("brand", BRAND_CASES)
+    def test_get_model_info(self, brand):
+        model_info = litellm.get_model_info(f"{brand['provider']}/qwen-max")
+        dashscope_info = litellm.get_model_info("dashscope/qwen-max")
+        assert model_info["litellm_provider"] == brand["provider"]
+        assert model_info["input_cost_per_token"] == dashscope_info["input_cost_per_token"]
+        assert model_info["output_cost_per_token"] == dashscope_info["output_cost_per_token"]
+
+    @pytest.mark.parametrize("brand", BRAND_CASES)
+    def test_flat_pricing_matches_dashscope(self, brand):
+        usage = Usage(prompt_tokens=1000, completion_tokens=500)
+        brand_costs = dashscope_cost_per_token(model="qwen-max", usage=usage, custom_llm_provider=brand["provider"])
+        dashscope_costs = dashscope_cost_per_token(model="qwen-max", usage=usage)
+        assert brand_costs == dashscope_costs
+
+    @pytest.mark.parametrize("brand", BRAND_CASES)
+    def test_tiered_pricing_matches_dashscope(self, brand):
+        usage = Usage(prompt_tokens=300000, completion_tokens=300000)
+        brand_costs = dashscope_cost_per_token(model="qwen-flash", usage=usage, custom_llm_provider=brand["provider"])
+        dashscope_costs = dashscope_cost_per_token(model="qwen-flash", usage=usage)
+        assert brand_costs == dashscope_costs
+        tier_2 = litellm.get_model_info(f"{brand['provider']}/qwen-flash")["tiered_pricing"][1]
+        assert math.isclose(brand_costs[0], 300000 * tier_2["input_cost_per_token"], rel_tol=1e-10)
+
+    @pytest.mark.parametrize("brand", BRAND_CASES)
+    def test_public_cost_per_token_routes_to_dashscope_calculator(self, brand):
+        brand_costs = litellm.cost_per_token(
+            model=f"{brand['provider']}/qwen-max",
+            prompt_tokens=1000,
+            completion_tokens=500,
+            custom_llm_provider=brand["provider"],
+        )
+        dashscope_costs = litellm.cost_per_token(
+            model="dashscope/qwen-max",
+            prompt_tokens=1000,
+            completion_tokens=500,
+            custom_llm_provider="dashscope",
+        )
+        assert brand_costs == dashscope_costs
+
+
+class TestQwenBrandCompletionMock:
+    @pytest.mark.respx()
+    @pytest.mark.parametrize("brand", BRAND_CASES)
+    def test_completion_hits_brand_default_host(self, respx_mock, brand, monkeypatch):
+        monkeypatch.setattr(litellm, "disable_aiohttp_transport", True)
+        respx_mock.post(f"{brand['default_base']}/chat/completions").respond(
+            json={
+                "id": "chatcmpl-123",
+                "object": "chat.completion",
+                "created": 1677652288,
+                "model": "qwen-turbo",
+                "choices": [
+                    {
+                        "index": 0,
+                        "message": {"role": "assistant", "content": "Hey from LiteLLM!"},
+                        "finish_reason": "stop",
+                    }
+                ],
+                "usage": {
+                    "prompt_tokens": 9,
+                    "completion_tokens": 12,
+                    "total_tokens": 21,
+                },
+            },
+            status_code=200,
+        )
+
+        response = completion(
+            model=f"{brand['provider']}/qwen-turbo",
+            messages=[{"role": "user", "content": "say hey from LiteLLM"}],
+            api_key="fake-brand-key",
+        )
+
+        assert response.choices[0].message.content == "Hey from LiteLLM!"
+        request = respx_mock.calls[0].request
+        assert request.url == f"{brand['default_base']}/chat/completions"
+        assert request.headers["Authorization"] == "Bearer fake-brand-key"

tests/test_litellm/proxy/test_proxy_cli.py

@@ -1737,7 +1737,8 @@ class TestRunServerDbSetup:
         mock_atexit_register,
         mock_subprocess_run,
     ):
-        """Test that use_prisma_db_push flag correctly controls PrismaManager.setup_database use_migrate parameter"""
+        """Which resolver and which migration mode run_server hands setup_database,
+        across the db push flag, the v2/legacy flag pair and USE_V2_MIGRATION_RESOLVER."""
         from litellm.proxy.proxy_cli import run_server
 
         # Mock subprocess.run to simulate prisma being available
@@ -1787,7 +1788,7 @@ class TestRunServerDbSetup:
             # use_prisma_db_push should be False (default), so use_migrate should be True
             run_server.main(["--local", "--skip_server_startup"], standalone_mode=False)
             mock_setup_database.assert_called_with(
-                use_migrate=True, use_v2_resolver=False
+                use_migrate=True, use_v2_resolver=True
             )
 
             # Reset mocks
@@ -1802,9 +1803,38 @@ class TestRunServerDbSetup:
                 standalone_mode=False,
             )
             mock_setup_database.assert_called_with(
-                use_migrate=False, use_v2_resolver=False
+                use_migrate=False, use_v2_resolver=True
             )
 
+            for argv, env_value, expected_v2 in (
+                ([], None, True),
+                (["--use_v2_migration_resolver"], None, True),
+                (["--use_legacy_migration_resolver"], None, False),
+                ([], "false", False),
+                ([], "true", True),
+                (["--use_v2_migration_resolver"], "false", True),
+                (["--use_legacy_migration_resolver"], "true", False),
+            ):
+                mock_setup_database.reset_mock()
+                mock_should_update_schema.reset_mock()
+                mock_should_update_schema.return_value = True
+
+                resolver_env = (
+                    {"USE_V2_MIGRATION_RESOLVER": env_value}
+                    if env_value is not None
+                    else {}
+                )
+                os.environ.pop("USE_V2_MIGRATION_RESOLVER", None)
+                with patch.dict(os.environ, resolver_env):
+                    run_server.main(
+                        ["--local", "--skip_server_startup", *argv],
+                        standalone_mode=False,
+                    )
+                assert mock_setup_database.call_args.kwargs == {
+                    "use_migrate": True,
+                    "use_v2_resolver": expected_v2,
+                }, f"argv={argv} env={env_value}"
+
     @patch("subprocess.run")
     @patch("atexit.register")
     @patch("litellm.proxy.db.prisma_client.PrismaManager.setup_database")
@@ -1869,7 +1899,7 @@ class TestRunServerDbSetup:
                 )
             assert exc_info.value.code == 1
             mock_setup_database.assert_called_once_with(
-                use_migrate=True, use_v2_resolver=False
+                use_migrate=True, use_v2_resolver=True
             )
 
     @patch("subprocess.run")
@@ -1981,7 +2011,6 @@ class TestRunServerDbSetup:
             use_migrate=True, use_v2_resolver=True
         )
 
-
 # --- Module-level helpers for worker startup hook tests ---
 
 _dummy_hook_called = False

ui/litellm-dashboard/package-lock.json

@@ -4891,9 +4891,9 @@
       }
     },
     "node_modules/baseline-browser-mapping": {
-      "version": "2.10.27",
-      "resolved": "https://registry.npmjs.org/baseline-browser-mapping/-/baseline-browser-mapping-2.10.27.tgz",
-      "integrity": "sha512-zEs/ufmZoUd7WftKpKyXaT6RFxpQ5Qm9xytKRHvJfxFV9DFJkZph9RvJ1LcOUi0Z1ZVijMte65JbILeV+8QQEA==",
+      "version": "2.11.20",
+      "resolved": "https://registry.npmjs.org/baseline-browser-mapping/-/baseline-browser-mapping-2.11.20.tgz",
+      "integrity": "sha512-H0ulySigv6icDJ1F7SjtdCD6PrhTpdYCmP0CactWy1+ekh0AFd0o1Wn5T8b+hnTmdBx19u9yhL6wvCylXMY7zw==",
       "license": "Apache-2.0",
       "bin": {
         "baseline-browser-mapping": "dist/cli.cjs"
@@ -4939,9 +4939,9 @@
       }
     },
     "node_modules/browserslist": {
-      "version": "4.28.2",
-      "resolved": "https://registry.npmjs.org/browserslist/-/browserslist-4.28.2.tgz",
-      "integrity": "sha512-48xSriZYYg+8qXna9kwqjIVzuQxi+KYWp2+5nCYnYKPTr0LvD89Jqk2Or5ogxz0NUMfIjhh2lIUX/LyX9B4oIg==",
+      "version": "4.28.8",
+      "resolved": "https://registry.npmjs.org/browserslist/-/browserslist-4.28.8.tgz",
+      "integrity": "sha512-V2NpofLblG64mfOtSgDhOJESZEGogzDMBv/q+W6oc4LXWP/q75eOXoOaaOu1EOadB9U4Bwx/e0yzbvwKH8zalA==",
       "dev": true,
       "funding": [
         {
@@ -4959,11 +4959,11 @@
       ],
       "license": "MIT",
       "dependencies": {
-        "baseline-browser-mapping": "^2.10.12",
-        "caniuse-lite": "^1.0.30001782",
-        "electron-to-chromium": "^1.5.328",
-        "node-releases": "^2.0.36",
-        "update-browserslist-db": "^1.2.3"
+        "baseline-browser-mapping": "^2.11.12",
+        "caniuse-lite": "^1.0.30001809",
+        "electron-to-chromium": "^1.5.402",
+        "node-releases": "^2.0.53",
+        "update-browserslist-db": "^1.3.0"
       },
       "bin": {
         "browserslist": "cli.js"
@@ -5042,9 +5042,9 @@
       }
     },
     "node_modules/caniuse-lite": {
-      "version": "1.0.30001791",
-      "resolved": "https://registry.npmjs.org/caniuse-lite/-/caniuse-lite-1.0.30001791.tgz",
-      "integrity": "sha512-yk0l/YSrOnFZk3UROpDLQD9+kC1l4meK/wed583AXrzoarMGJcbRi2Q4RaUYbKxYAsZ8sWmaSa/DsLmdBeI1vQ==",
+      "version": "1.0.30001810",
+      "resolved": "https://registry.npmjs.org/caniuse-lite/-/caniuse-lite-1.0.30001810.tgz",
+      "integrity": "sha512-TITQPUkaz+aVk5GL6NhOdwk1aEaNTSDPsGFWrTuhKGtjTF70jL/Oht2W4c6rXUe5fu7Ie19VIahAXHIIiWWNeg==",
       "funding": [
         {
           "type": "opencollective",
@@ -5706,9 +5706,9 @@
       }
     },
     "node_modules/electron-to-chromium": {
-      "version": "1.5.349",
-      "resolved": "https://registry.npmjs.org/electron-to-chromium/-/electron-to-chromium-1.5.349.tgz",
-      "integrity": "sha512-QsWVGyRuY07Aqb234QytTfwd5d9AJlfNIQ5wIOl1L+PZDzI9d9+Fn0FRale/QYlFxt/bUnB0/nLd1jFPGxGK1A==",
+      "version": "1.5.416",
+      "resolved": "https://registry.npmjs.org/electron-to-chromium/-/electron-to-chromium-1.5.416.tgz",
+      "integrity": "sha512-K6bvB2BjnNrugtIih6ewlbBI9DXa976jIdiIlRLHhBoEI9a4JaQjjHyF+A1IQI543aQYR4LnmOrT/K5fZj0aPA==",
       "dev": true,
       "license": "ISC"
     },
@@ -9901,11 +9901,14 @@
       }
     },
     "node_modules/node-releases": {
-      "version": "2.0.38",
-      "resolved": "https://registry.npmjs.org/node-releases/-/node-releases-2.0.38.tgz",
-      "integrity": "sha512-3qT/88Y3FbH/Kx4szpQQ4HzUbVrHPKTLVpVocKiLfoYvw9XSGOX2FmD2d6DrXbVYyAQTF2HeF6My8jmzx7/CRw==",
+      "version": "2.0.54",
+      "resolved": "https://registry.npmjs.org/node-releases/-/node-releases-2.0.54.tgz",
+      "integrity": "sha512-YHs7BmmcsdAI5Ozuf8JZo6PT0mv2GIWC9vMfvUC3dp65M8hn7Ux8CPL+2oBI7juNuj9d0ndhTcznq2ODBps9cQ==",
       "dev": true,
-      "license": "MIT"
+      "license": "MIT",
+      "engines": {
+        "node": ">=18"
+      }
     },
     "node_modules/nuqs": {
       "version": "2.9.4",
@@ -12350,9 +12353,9 @@
       }
     },
     "node_modules/update-browserslist-db": {
-      "version": "1.2.3",
-      "resolved": "https://registry.npmjs.org/update-browserslist-db/-/update-browserslist-db-1.2.3.tgz",
-      "integrity": "sha512-Js0m9cx+qOgDxo0eMiFGEueWztz+d4+M3rGlmKPT+T4IS/jP4ylw3Nwpu6cpTTP8R1MAC1kF4VbdLt3ARf209w==",
+      "version": "1.3.2",
+      "resolved": "https://registry.npmjs.org/update-browserslist-db/-/update-browserslist-db-1.3.2.tgz",
+      "integrity": "sha512-UQ+MSxlhRm1bzjhU+DcuXfjFO1FzNtqhK5+9Yvlp90ItDLk5vT932A0rFu619nf7RVS+Y/VeaUW1jaRDqZ8VJw==",
       "dev": true,
       "funding": [
         {

ui/litellm-dashboard/src/components/CreateUserButton.test.tsx

@@ -5,6 +5,7 @@ import { describe, it, expect, vi, beforeEach } from "vitest";
 import { CreateUserButton } from "./CreateUserButton";
 import * as networking from "./networking";
 import { toast } from "@/lib/toast";
+import { expectControlBesideLabel } from "../../tests/fieldOrientation";
 
 vi.mock("./networking", () => ({
   userCreateCall: vi.fn(),
@@ -294,6 +295,20 @@ describe("CreateUserButton", () => {
     });
   });
 
+  it("lays the send invitation email checkbox out beside its label", async () => {
+    const user = userEvent.setup({ pointerEventsCheck: PointerEventsCheckLevel.Never });
+    renderWithProviders(<CreateUserButton {...defaultProps} />);
+
+    await waitFor(() => {
+      expect(screen.getByRole("button", { name: /\+ invite user/i })).toBeInTheDocument();
+    });
+    await user.click(screen.getByRole("button", { name: /\+ invite user/i }));
+
+    const dialog = screen.getByRole("dialog", { name: /invite user/i });
+
+    expectControlBesideLabel(within(dialog).getByRole("checkbox"));
+  });
+
   describe("organizations", () => {
     it("should send organizations list in POST body when organizations are selected", async () => {
       const { useOrganizations } = await import("@/app/(dashboard)/hooks/organizations/useOrganizations");

ui/litellm-dashboard/src/components/CreateUserButton.tsx

@@ -270,7 +270,7 @@ export const CreateUserButton: React.FC<CreateuserProps> = ({
   );
 
   const sendInviteEmailField = (
-    <FormField control={form.control} name="send_invite_email" label="Send invitation email">
+    <FormField control={form.control} name="send_invite_email" label="Send invitation email" orientation="horizontal">
       {({ id, value, onChange, onBlur }) => (
         <Checkbox id={id} checked={value} onCheckedChange={onChange} onBlur={onBlur} />
       )}

ui/litellm-dashboard/src/components/Settings/AdminSettings/SSOSettings/Modals/BaseSSOSettingsForm.test.tsx

@@ -9,6 +9,7 @@ import BaseSSOSettingsForm, {
   submitMountedSSOValues,
   useSSOSettingsForm,
 } from "./BaseSSOSettingsForm";
+import { expectControlBesideLabel } from "../../../../../../tests/fieldOrientation";
 
 const user = () => userEvent.setup({ pointerEventsCheck: 0 });
 
@@ -233,6 +234,38 @@ describe("BaseSSOSettingsForm", () => {
 
     expect(screen.queryByText("Use Team Mappings")).not.toBeInTheDocument();
   });
+
+  it("lays a provider checkbox field out beside its label", async () => {
+    const TestWrapper = () => {
+      const form = useSSOSettingsForm("sso-settings");
+
+      return <BaseSSOSettingsForm form={form} onFormSubmit={vi.fn()} />;
+    };
+
+    renderWithProviders(<TestWrapper />);
+
+    await openProviderDropdown();
+    await user().click(await screen.findByText(/saml sso/i));
+
+    expectControlBesideLabel(
+      await screen.findByRole("checkbox", { name: "Allow IdP-initiated (unsolicited) responses" }),
+    );
+  });
+
+  it.each(["Use Role Mappings", "Use Team Mappings"])("lays the %s toggle out beside its label", async (label) => {
+    const TestWrapper = () => {
+      const form = useSSOSettingsForm("sso-settings");
+
+      return <BaseSSOSettingsForm form={form} onFormSubmit={vi.fn()} />;
+    };
+
+    renderWithProviders(<TestWrapper />);
+
+    await openProviderDropdown();
+    await user().click(await screen.findByText(/okta/i));
+
+    expectControlBesideLabel(await screen.findByRole("checkbox", { name: label }));
+  });
 });
 
 describe("renderProviderFields", () => {

ui/litellm-dashboard/src/components/Settings/AdminSettings/SSOSettings/Modals/BaseSSOSettingsForm.tsx

@@ -303,7 +303,7 @@ const SSOProviderField = ({ field }: { field: SSOProviderConfig["fields"][number
 
   if (field.type === "checkbox") {
     return (
-      <FormField control={control} name={field.name} label={field.label}>
+      <FormField control={control} name={field.name} label={field.label} orientation="horizontal">
         {({ value, onChange, onBlur, id, ...rest }) => (
           <Checkbox
             id={id}
@@ -413,7 +413,7 @@ export const MappingToggleField = ({
   const { control } = useFormContext<SSOSettingsFormValues>();
 
   return (
-    <FormField control={control} name={name} label={label}>
+    <FormField control={control} name={name} label={label} orientation="horizontal">
       {({ value, onChange, onBlur, id, ...rest }) => (
         <Checkbox
           id={id}

ui/litellm-dashboard/src/components/claude_code_plugins/skill_detail.tsx

@@ -1,5 +1,6 @@
 import React, { useState } from "react";
 import { ArrowLeft, Check, Copy, Link2 } from "lucide-react";
+import { cn } from "@/lib/cva.config";
 import { buildMarketplaceSettingsSnippet, formatInstallCommand } from "./helpers";
 import { Plugin } from "./types";
 
@@ -50,48 +51,37 @@ const SkillDetail: React.FC<SkillDetailProps> = ({ skill, onBack }) => {
   ];
 
   return (
-    <div style={{ padding: "24px 32px 24px 0" }}>
+    <div className="py-6 pl-0 pr-8">
       {/* Back link */}
       <div
         onClick={onBack}
-        style={{
-          display: "inline-flex",
-          alignItems: "center",
-          gap: 6,
-          color: "#5f6368",
-          cursor: "pointer",
-          fontSize: 14,
-          marginBottom: 24,
-        }}
+        className="mb-6 inline-flex cursor-pointer items-center gap-1.5 text-sm text-muted-foreground"
       >
         <ArrowLeft className="size-3" />
         <span>Skills</span>
       </div>
 
       {/* Header */}
-      <div style={{ marginBottom: 8 }}>
-        <h1 style={{ fontSize: 28, fontWeight: 400, color: "#202124", margin: 0, lineHeight: 1.2 }}>{skill.name}</h1>
+      <div className="mb-2">
+        <h1 className="m-0 text-[28px] font-normal leading-tight text-foreground">{skill.name}</h1>
         {skill.description && (
-          <p style={{ fontSize: 14, color: "#5f6368", margin: "8px 0 0 0", lineHeight: 1.6 }}>{skill.description}</p>
+          <p className="mb-0 ml-0 mr-0 mt-2 text-sm leading-relaxed text-muted-foreground">{skill.description}</p>
         )}
       </div>
 
       {/* Tab bar */}
-      <div style={{ borderBottom: "1px solid #dadce0", marginBottom: 28, marginTop: 24 }}>
-        <div style={{ display: "flex", gap: 0 }}>
+      <div className="mb-7 mt-6 border-b border-border">
+        <div className="flex">
           {tabs.map((tab) => (
             <div
               key={tab.key}
               onClick={() => setActiveTab(tab.key)}
-              style={{
-                padding: "12px 20px",
-                fontSize: 14,
-                color: activeTab === tab.key ? "#1a73e8" : "#5f6368",
-                borderBottom: activeTab === tab.key ? "3px solid #1a73e8" : "3px solid transparent",
-                cursor: "pointer",
-                fontWeight: activeTab === tab.key ? 500 : 400,
-                marginBottom: -1,
-              }}
+              className={cn(
+                "-mb-px cursor-pointer border-b-[3px] px-5 py-3 text-sm",
+                activeTab === tab.key
+                  ? "border-info font-medium text-info"
+                  : "border-transparent font-normal text-muted-foreground",
+              )}
             >
               {tab.label}
             </div>
@@ -101,27 +91,23 @@ const SkillDetail: React.FC<SkillDetailProps> = ({ skill, onBack }) => {
 
       {/* Overview tab */}
       {activeTab === "overview" && (
-        <div style={{ display: "flex", gap: 64 }}>
+        <div className="flex gap-16">
           {/* Left column */}
-          <div style={{ flex: 1, minWidth: 0 }}>
-            <h2 style={{ fontSize: 18, fontWeight: 400, color: "#202124", margin: "0 0 4px 0" }}>Skill Details</h2>
-            <p style={{ fontSize: 13, color: "#5f6368", margin: "0 0 16px 0" }}>Metadata registered with this skill</p>
-            <table style={{ width: "100%", borderCollapse: "collapse", fontSize: 14 }}>
+          <div className="min-w-0 flex-1">
+            <h2 className="m-0 mb-1 text-lg font-normal text-foreground">Skill Details</h2>
+            <p className="m-0 mb-4 text-[13px] text-muted-foreground">Metadata registered with this skill</p>
+            <table className="w-full border-collapse text-sm">
               <thead>
-                <tr style={{ borderBottom: "1px solid #dadce0" }}>
-                  <th style={{ textAlign: "left", padding: "12px 0", color: "#5f6368", fontWeight: 500, width: 160 }}>
-                    Property
-                  </th>
-                  <th style={{ textAlign: "left", padding: "12px 0", color: "#5f6368", fontWeight: 500 }}>
-                    {skill.name}
-                  </th>
+                <tr className="border-b border-border">
+                  <th className="w-40 py-3 text-left font-medium text-muted-foreground">Property</th>
+                  <th className="py-3 text-left font-medium text-muted-foreground">{skill.name}</th>
                 </tr>
               </thead>
               <tbody>
                 {detailRows.map((row, i) => (
-                  <tr key={i} style={{ borderBottom: "1px solid #f1f3f4" }}>
-                    <td style={{ padding: "12px 0", color: "#3c4043" }}>{row.property}</td>
-                    <td style={{ padding: "12px 0", color: "#202124" }}>{row.value}</td>
+                  <tr key={i} className="border-b border-border">
+                    <td className="py-3 text-foreground">{row.property}</td>
+                    <td className="py-3 text-foreground">{row.value}</td>
                   </tr>
                 ))}
               </tbody>
@@ -129,38 +115,27 @@ const SkillDetail: React.FC<SkillDetailProps> = ({ skill, onBack }) => {
           </div>
 
           {/* Right sidebar */}
-          <div style={{ width: 240, flexShrink: 0 }}>
-            <div style={{ marginBottom: 24 }}>
-              <div style={{ fontSize: 12, color: "#5f6368", marginBottom: 4 }}>Status</div>
+          <div className="w-60 shrink-0">
+            <div className="mb-6">
+              <div className="mb-1 text-xs text-muted-foreground">Status</div>
               <span
-                style={{
-                  fontSize: 12,
-                  padding: "3px 10px",
-                  borderRadius: 12,
-                  backgroundColor: skill.enabled ? "#e6f4ea" : "#f1f3f4",
-                  color: skill.enabled ? "#137333" : "#5f6368",
-                  fontWeight: 500,
-                }}
+                className={cn(
+                  "rounded-xl px-2.5 py-[3px] text-xs font-medium",
+                  skill.enabled ? "bg-success/10 text-success" : "bg-muted text-muted-foreground",
+                )}
               >
                 {skill.enabled ? "Public" : "Draft"}
               </span>
             </div>
 
             {sourceUrl && (
-              <div style={{ marginBottom: 24 }}>
-                <div style={{ fontSize: 12, color: "#5f6368", marginBottom: 4 }}>Source</div>
+              <div className="mb-6">
+                <div className="mb-1 text-xs text-muted-foreground">Source</div>
                 <a
                   href={sourceUrl}
                   target="_blank"
                   rel="noopener noreferrer"
-                  style={{
-                    fontSize: 13,
-                    color: "#1a73e8",
-                    wordBreak: "break-all",
-                    display: "flex",
-                    alignItems: "center",
-                    gap: 4,
-                  }}
+                  className="flex items-center gap-1 break-all text-[13px] text-info"
                 >
                   {sourceUrl.replace("https://", "")}
                   <Link2 className="size-3 shrink-0" />
@@ -169,20 +144,13 @@ const SkillDetail: React.FC<SkillDetailProps> = ({ skill, onBack }) => {
             )}
 
             {skill.keywords && skill.keywords.length > 0 && (
-              <div style={{ marginBottom: 24 }}>
-                <div style={{ fontSize: 12, color: "#5f6368", marginBottom: 8 }}>Tags</div>
-                <div style={{ display: "flex", flexWrap: "wrap", gap: 6 }}>
+              <div className="mb-6">
+                <div className="mb-2 text-xs text-muted-foreground">Tags</div>
+                <div className="flex flex-wrap gap-1.5">
                   {skill.keywords.map((kw) => (
                     <span
                       key={kw}
-                      style={{
-                        fontSize: 12,
-                        padding: "4px 12px",
-                        borderRadius: 16,
-                        border: "1px solid #dadce0",
-                        color: "#3c4043",
-                        backgroundColor: "#fff",
-                      }}
+                      className="rounded-2xl border border-border bg-card px-3 py-1 text-xs text-foreground"
                     >
                       {kw}
                     </span>
@@ -192,10 +160,8 @@ const SkillDetail: React.FC<SkillDetailProps> = ({ skill, onBack }) => {
             )}
 
             <div>
-              <div style={{ fontSize: 12, color: "#5f6368", marginBottom: 4 }}>Skill ID</div>
-              <div style={{ fontSize: 12, fontFamily: "monospace", color: "#3c4043", wordBreak: "break-all" }}>
-                {skill.id}
-              </div>
+              <div className="mb-1 text-xs text-muted-foreground">Skill ID</div>
+              <div className="break-all font-mono text-xs text-foreground">{skill.id}</div>
             </div>
           </div>
         </div>
@@ -203,93 +169,43 @@ const SkillDetail: React.FC<SkillDetailProps> = ({ skill, onBack }) => {
 
       {/* How to Use tab */}
       {activeTab === "usage" && (
-        <div style={{ maxWidth: 640 }}>
-          <h2 style={{ fontSize: 18, fontWeight: 400, color: "#202124", margin: "0 0 8px 0" }}>Using this skill</h2>
-          <p style={{ fontSize: 14, color: "#5f6368", margin: "0 0 24px 0", lineHeight: 1.6 }}>
+        <div className="max-w-[640px]">
+          <h2 className="m-0 mb-2 text-lg font-normal text-foreground">Using this skill</h2>
+          <p className="m-0 mb-6 text-sm leading-relaxed text-muted-foreground">
             Once your proxy is set as a marketplace, enable this skill in Claude Code with one command:
           </p>
 
           {/* Install command */}
-          <div
-            style={{
-              border: "1px solid #dadce0",
-              borderRadius: 8,
-              overflow: "hidden",
-              marginBottom: 24,
-            }}
-          >
-            <div
-              style={{
-                display: "flex",
-                alignItems: "center",
-                justifyContent: "space-between",
-                padding: "10px 16px",
-                backgroundColor: "#f8f9fa",
-                borderBottom: "1px solid #dadce0",
-              }}
-            >
-              <span style={{ fontSize: 13, color: "#3c4043", fontWeight: 500 }}>Run in Claude Code</span>
+          <div className="mb-6 overflow-hidden rounded-lg border border-border">
+            <div className="flex items-center justify-between border-b border-border bg-muted px-4 py-2.5">
+              <span className="text-[13px] font-medium text-foreground">Run in Claude Code</span>
               <button
                 onClick={() => copyToClipboard(installCommand, "install")}
-                style={{
-                  display: "flex",
-                  alignItems: "center",
-                  gap: 4,
-                  fontSize: 12,
-                  color: copiedKey === "install" ? "#137333" : "#1a73e8",
-                  background: "none",
-                  border: "none",
-                  cursor: "pointer",
-                  padding: 0,
-                }}
+                className={cn(
+                  "flex cursor-pointer items-center gap-1 border-none bg-transparent p-0 text-xs",
+                  copiedKey === "install" ? "text-success" : "text-info",
+                )}
               >
                 {copiedKey === "install" ? <Check className="size-3" /> : <Copy className="size-3" />}
                 {copiedKey === "install" ? "Copied" : "Copy"}
               </button>
             </div>
-            <pre
-              style={{
-                margin: 0,
-                padding: "14px 16px",
-                fontSize: 14,
-                fontFamily: "monospace",
-                color: "#202124",
-                backgroundColor: "#fff",
-              }}
-            >
-              {installCommand}
-            </pre>
+            <pre className="m-0 bg-card px-4 py-3.5 font-mono text-sm text-foreground">{installCommand}</pre>
           </div>
 
           {/* Shown when the marketplace catalog is stale and the plugin isn't found yet */}
-          <div
-            style={{
-              border: "1px solid #fce8b2",
-              borderRadius: 8,
-              padding: "12px 16px",
-              backgroundColor: "#fefce8",
-              marginBottom: 16,
-            }}
-          >
-            <p style={{ fontSize: 13, color: "#5f6368", lineHeight: 1.6, margin: "0 0 8px 0" }}>
+          <div className="mb-4 rounded-lg border border-warning/30 bg-warning/10 px-4 py-3">
+            <p className="m-0 mb-2 text-[13px] leading-relaxed text-muted-foreground">
               If you see &quot;Plugin {skill.name} not found in marketplace&quot;, update the catalog first:
             </p>
-            <pre
-              style={{
-                margin: 0,
-                fontSize: 13,
-                fontFamily: "monospace",
-                color: "#202124",
-                backgroundColor: "transparent",
-              }}
-            >
+            <pre className="m-0 bg-transparent font-mono text-[13px] text-foreground">
               /plugin marketplace update litellm
             </pre>
           </div>
 
-          <p style={{ fontSize: 13, color: "#5f6368", lineHeight: 1.6, margin: 0 }}>
+          <p className="m-0 text-[13px] leading-relaxed text-muted-foreground">
             Don&apos;t have the marketplace configured yet?{" "}
-            <span onClick={() => setActiveTab("setup")} style={{ color: "#1a73e8", cursor: "pointer" }}>
+            <span onClick={() => setActiveTab("setup")} className="cursor-pointer text-info">
               See one-time setup →
             </span>
           </p>
@@ -298,126 +214,56 @@ const SkillDetail: React.FC<SkillDetailProps> = ({ skill, onBack }) => {
 
       {/* Setup tab (linked from usage) */}
       {activeTab === "setup" && (
-        <div style={{ maxWidth: 640 }}>
-          <h2 style={{ fontSize: 18, fontWeight: 400, color: "#202124", margin: "0 0 8px 0" }}>
-            One-time marketplace setup
-          </h2>
+        <div className="max-w-[640px]">
+          <h2 className="m-0 mb-2 text-lg font-normal text-foreground">One-time marketplace setup</h2>
 
           {/* Option 1: single command — fastest path for most users */}
-          <p style={{ fontSize: 14, color: "#5f6368", margin: "0 0 12px 0", lineHeight: 1.6 }}>
+          <p className="m-0 mb-3 text-sm leading-relaxed text-muted-foreground">
             Run this command in Claude Code to register the marketplace:
           </p>
-          <div
-            style={{
-              border: "1px solid #dadce0",
-              borderRadius: 8,
-              overflow: "hidden",
-              marginBottom: 24,
-            }}
-          >
-            <div
-              style={{
-                display: "flex",
-                alignItems: "center",
-                justifyContent: "space-between",
-                padding: "10px 16px",
-                backgroundColor: "#f8f9fa",
-                borderBottom: "1px solid #dadce0",
-              }}
-            >
-              <span style={{ fontSize: 13, color: "#3c4043", fontWeight: 500 }}>Run in Claude Code</span>
+          <div className="mb-6 overflow-hidden rounded-lg border border-border">
+            <div className="flex items-center justify-between border-b border-border bg-muted px-4 py-2.5">
+              <span className="text-[13px] font-medium text-foreground">Run in Claude Code</span>
               <button
                 onClick={() => {
                   const origin = typeof window !== "undefined" ? window.location.origin : "";
                   copyToClipboard(`/plugin marketplace add ${origin}/claude-code/marketplace.json`, "marketplace-cmd");
                 }}
-                style={{
-                  display: "flex",
-                  alignItems: "center",
-                  gap: 4,
-                  fontSize: 12,
-                  color: copiedKey === "marketplace-cmd" ? "#137333" : "#1a73e8",
-                  background: "none",
-                  border: "none",
-                  cursor: "pointer",
-                  padding: 0,
-                }}
+                className={cn(
+                  "flex cursor-pointer items-center gap-1 border-none bg-transparent p-0 text-xs",
+                  copiedKey === "marketplace-cmd" ? "text-success" : "text-info",
+                )}
               >
                 {copiedKey === "marketplace-cmd" ? <Check className="size-3" /> : <Copy className="size-3" />}
                 {copiedKey === "marketplace-cmd" ? "Copied" : "Copy"}
               </button>
             </div>
-            <pre
-              style={{
-                margin: 0,
-                padding: "14px 16px",
-                fontSize: 13,
-                fontFamily: "monospace",
-                color: "#202124",
-                backgroundColor: "#fff",
-              }}
-            >
+            <pre className="m-0 bg-card px-4 py-3.5 font-mono text-[13px] text-foreground">
               {`/plugin marketplace add ${typeof window !== "undefined" ? window.location.origin : "<proxy-url>"}/claude-code/marketplace.json`}
             </pre>
           </div>
 
           {/* Option 2: settings.json — for persistent config or managed deployments.
               extraKnownMarketplaces requires source to be a nested object, not a flat string. */}
-          <p style={{ fontSize: 14, color: "#5f6368", margin: "0 0 12px 0", lineHeight: 1.6 }}>
-            Or add this to{" "}
-            <code style={{ fontSize: 13, backgroundColor: "#f1f3f4", padding: "1px 6px", borderRadius: 4 }}>
-              ~/.claude/settings.json
-            </code>{" "}
+          <p className="m-0 mb-3 text-sm leading-relaxed text-muted-foreground">
+            Or add this to <code className="rounded bg-muted px-1.5 py-px text-[13px]">~/.claude/settings.json</code>{" "}
             for a persistent configuration:
           </p>
-          <div
-            style={{
-              border: "1px solid #dadce0",
-              borderRadius: 8,
-              overflow: "hidden",
-            }}
-          >
-            <div
-              style={{
-                display: "flex",
-                alignItems: "center",
-                justifyContent: "space-between",
-                padding: "10px 16px",
-                backgroundColor: "#f8f9fa",
-                borderBottom: "1px solid #dadce0",
-              }}
-            >
-              <span style={{ fontSize: 13, color: "#3c4043", fontWeight: 500 }}>~/.claude/settings.json</span>
+          <div className="overflow-hidden rounded-lg border border-border">
+            <div className="flex items-center justify-between border-b border-border bg-muted px-4 py-2.5">
+              <span className="text-[13px] font-medium text-foreground">~/.claude/settings.json</span>
               <button
                 onClick={() => copyToClipboard(settingsSnippet, "settings")}
-                style={{
-                  display: "flex",
-                  alignItems: "center",
-                  gap: 4,
-                  fontSize: 12,
-                  color: copiedKey === "settings" ? "#137333" : "#1a73e8",
-                  background: "none",
-                  border: "none",
-                  cursor: "pointer",
-                  padding: 0,
-                }}
+                className={cn(
+                  "flex cursor-pointer items-center gap-1 border-none bg-transparent p-0 text-xs",
+                  copiedKey === "settings" ? "text-success" : "text-info",
+                )}
               >
                 {copiedKey === "settings" ? <Check className="size-3" /> : <Copy className="size-3" />}
                 {copiedKey === "settings" ? "Copied" : "Copy"}
               </button>
             </div>
-            <pre
-              style={{
-                margin: 0,
-                padding: "14px 16px",
-                fontSize: 13,
-                fontFamily: "monospace",
-                color: "#202124",
-                backgroundColor: "#fff",
-              }}
-            >
-              {settingsSnippet}
-            </pre>
+            <pre className="m-0 bg-card px-4 py-3.5 font-mono text-[13px] text-foreground">{settingsSnippet}</pre>
           </div>
         </div>
       )}

ui/litellm-dashboard/src/components/provider_info_helpers.tsx

@@ -150,6 +150,8 @@ export enum Providers {
   PETALS = "Petals",
   PG_VECTOR = "Pg Vector",
   PREDIBASE = "Predibase",
+  Qwen_AI_Platform = "Qwen AI Platform",
+  QwenCloud = "QwenCloud",
   RECRAFT = "Recraft",
   REPLICATE = "Replicate",
   RunwayML = "RunwayML",
@@ -262,6 +264,8 @@ export const provider_map: Record<string, string> = {
   PETALS: "petals",
   PG_VECTOR: "pg_vector",
   PREDIBASE: "predibase",
+  Qwen_AI_Platform: "qwen_ai_platform",
+  QwenCloud: "qwencloud",
   RECRAFT: "recraft",
   REPLICATE: "replicate",
   RunwayML: "runwayml",
@@ -357,6 +361,8 @@ export const providerLogoMap: Partial<Record<Providers, string>> = {
   [Providers.Openrouter]: openrouterLogo.src,
   [Providers.Oracle]: oracleLogo.src,
   [Providers.Perplexity]: perplexityAiLogo.src,
+  [Providers.Qwen_AI_Platform]: qwenLogo.src,
+  [Providers.QwenCloud]: qwenLogo.src,
   [Providers.RECRAFT]: recraftLogo.src,
   [Providers.REPLICATE]: replicateLogo.src,
   [Providers.RunwayML]: runwayLogo.src,

ui/litellm-dashboard/src/components/ui/field.test.tsx

@@ -14,6 +14,7 @@ import {
   FieldSet,
   FieldTitle,
 } from "./field";
+import { ROW_LAYOUT_CLASSES, STRETCH_CHILDREN_CLASS } from "../../../tests/fieldOrientation";
 
 describe("FieldError", () => {
   it("renders nothing when there are no errors and no children", () => {
@@ -85,6 +86,20 @@ describe("Field", () => {
 
     expect(screen.getByRole("group")).toHaveAttribute("data-orientation", "horizontal");
   });
+
+  it("stretches every child when vertical, which is what inputs, selects and textareas want", () => {
+    render(<Field />);
+
+    expect(screen.getByRole("group")).toHaveClass("flex-col", STRETCH_CHILDREN_CLASS);
+  });
+
+  it("lays children in a row at their own width when horizontal, so a checkbox stays square", () => {
+    render(<Field orientation="horizontal" />);
+    const field = screen.getByRole("group");
+
+    expect(field).toHaveClass(...ROW_LAYOUT_CLASSES);
+    expect(field).not.toHaveClass(STRETCH_CHILDREN_CLASS);
+  });
 });
 
 describe("field primitives forward refs to their DOM node", () => {

ui/litellm-dashboard/tests/fieldOrientation.ts

@@ -0,0 +1,18 @@
+import { expect } from "vitest";
+
+export const ROW_LAYOUT_CLASSES = ["flex-row", "items-center"] as const;
+export const STRETCH_CHILDREN_CLASS = "*:w-full";
+
+/**
+ * Asserts a control sits beside its label at its own width instead of being stretched across the
+ * field. Reaches for the resolved classes because the defect is purely visual: nothing accessible
+ * distinguishes a square checkbox from a full-width bar.
+ */
+export const expectControlBesideLabel = (control: HTMLElement): void => {
+  const field = control.closest('[data-slot="field"]');
+  if (field === null) throw new Error("control is not rendered inside a form field");
+
+  expect(field).toHaveAttribute("data-orientation", "horizontal");
+  expect(field).toHaveClass(...ROW_LAYOUT_CLASSES);
+  expect(field).not.toHaveClass(STRETCH_CHILDREN_CLASS);
+};

uv.lock

@@ -10,7 +10,7 @@ resolution-markers = [
 ]
 
 [options]
-exclude-newer = "2026-08-26T18:33:25.773031Z"
+exclude-newer = "2026-08-29T17:58:57.633306Z"
 exclude-newer-span = "P3D"
 
 [manifest]
@@ -4266,7 +4266,7 @@ wheels = [
 
 [[package]]
 name = "litellm"
-version = "1.100.0"
+version = "1.101.0"
 source = { editable = "." }
 dependencies = [
     { name = "aiohttp" },
@@ -4669,12 +4669,12 @@ proxy-dev = [
 
 [[package]]
 name = "litellm-enterprise"
-version = "0.1.62"
+version = "0.1.63"
 source = { editable = "enterprise" }
 
 [[package]]
 name = "litellm-proxy-extras"
-version = "0.4.91"
+version = "0.4.92"
 source = { editable = "litellm-proxy-extras" }
 
 [[package]]