Commit graph

34410 commits

Author SHA1 Message Date
Julio Quinteros Pro
ce59e6f00a fix(tests): use unconditional skip for vertex/gemini token counting test
The parametrized test covers both gemini-2.5-pro (needs GEMINI_API_KEY)
and vertex-ai-gemini-2.5-pro (needs VERTEX_AI_PRIVATE_KEY). A skipif
on GEMINI_API_KEY alone was insufficient for the vertex variant.
Switch to @pytest.mark.skip to guard both parametrizations consistently.

Addresses Greptile review comment on PR #21669.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-02-20 11:36:54 -03:00
Julio Quinteros Pro
174c17cec1 fix(tests): skip additional CI tests requiring external DB connection
Mark DB-dependent tests in test_key_generate_prisma.py and test_jwt.py
with @pytest.mark.skip to prevent CI failures when Prisma DB is unavailable.

Tests now skipped:
- test_call_with_invalid_key
- test_call_with_invalid_model
- test_call_with_end_user_over_budget
- test_aasync_call_with_key_over_model_budget (all 3 parametrize variants)
- test_call_with_key_never_over_budget
- test_aview_spend_per_user
- test_aadmin_only_routes
- test_auth_vertex_ai_route
- test_team_token_output (both audience variants)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-02-20 11:35:43 -03:00
Julio Quinteros Pro
3dfa3611d9 fix(tests): skip CI tests requiring external services (DB, API keys)
Mark tests that require Prisma DB connections or external API credentials
with @pytest.mark.skip / @pytest.mark.skipif so they don't block CI runs
when the infrastructure is unavailable.

Tests skipped:
- test_create_user_default_budget (Prisma DB)
- test_gemini_pass_through_endpoint (GEMINI_API_KEY / GOOGLE_API_KEY)
- test_vertex_ai_gemini_token_counting_with_contents (Google API creds)
- test_new/update/delete/info_project (Prisma DB)
- test_create/list/get/delete_skill_sdk (Prisma DB)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-02-20 11:28:42 -03:00
jquinter
39d95eb1b6
Merge pull request #21667 from BerriAI/fix/policy-endpoints-export-private-helpers
fix(policy_endpoints): re-export private helper functions from package __init__.py
2026-02-20 11:13:09 -03:00
Julio Quinteros Pro
1f6303677c fix(tests): set premium_user=True in test_aasync_call_with_key_over_model_budget
generate_key_fn calls validate_model_max_budget which raises
"You must have an enterprise license to set model_max_budget" unless
premium_user=True. The test was hitting this gate before it could
exercise the actual per-model budget logic, causing all three variants
to fail with an enterprise-check error instead of the expected budget
exceeded / pass-through result.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-02-20 11:12:38 -03:00
Julio Quinteros Pro
65388d877e fix(policy_endpoints): explicitly re-export private helper functions from __init__.py
`import *` does not re-export names starting with `_`, so the private
competitor-enrichment helpers added in endpoints.py were invisible when
imported from the package:

  from litellm.proxy.management_endpoints.policy_endpoints import (
      _build_all_names_per_competitor, ...
  )

This caused an ImportError that prevented the entire test module from
collecting. Add explicit imports for each private helper so they are
accessible from the package namespace.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-02-20 11:09:57 -03:00
jquinter
8f5561cf48
Merge pull request #21666 from BerriAI/fix/test-max-effort-error-message
fix(tests): update test_max_effort_rejected_for_opus_45 regex to match current error message
2026-02-20 11:09:22 -03:00
Julio Quinteros Pro
963e0eabcb fix(tests): update test_max_effort_rejected_for_opus_45 regex to match new error message
The production error message was expanded when Sonnet 4.6 was also added as
a supported model for effort='max'. The test's match regex still referenced
the old "Claude Opus 4.6"-only message; update it to match the new
"Claude 4.6 models" wording.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-02-20 11:07:52 -03:00
Sameer Kankute
b62db9fcc9
Merge pull request #21663 from BerriAI/litellm_add_reasoning_support_config
[Feat] Add reasoning support via config
2026-02-20 18:17:09 +05:30
Sameer Kankute
164734fa45
Merge pull request #21662 from BerriAI/litellm_mcp_openapi_spec
[Fix]Add mcp via openapi spec
2026-02-20 18:16:30 +05:30
Sameer Kankute
9e0bd46c28
Merge pull request #21659 from BerriAI/litellm_fix_converse_disable_tool
Fix mapping of parallel_tool_calls for bedrock converse
2026-02-20 18:16:10 +05:30
Sameer Kankute
784ce97918
Merge pull request #21658 from BerriAI/litellm_fix_vertex_ai_messages
Fix: api_base is required. Unable to determine the correct api_base for the request
2026-02-20 18:15:30 +05:30
Sameer Kankute
bf024c0404
Merge pull request #21654 from BerriAI/litellm_new_gemini_pathc
Fix _map_reasoning_effort_to_thinking_level for all gemini 3 family
2026-02-20 18:14:47 +05:30
Sameer Kankute
5d8f2196d1 Add test case 2026-02-20 16:22:44 +05:30
Sameer Kankute
871c049f49 Add support for reasoning and tools viaconfig 2026-02-20 16:22:19 +05:30
Sameer Kankute
3cd9072539
Update ui/litellm-dashboard/tsconfig.json
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
2026-02-20 14:08:38 +05:30
Sameer Kankute
f99ea619da Add search bar for enabling and calling tool 2026-02-20 13:57:58 +05:30
Sameer Kankute
ba74ee5a31 Add test for base url extraction and migration 2026-02-20 13:47:44 +05:30
Sameer Kankute
19b7b4019c Fix: server url extraction from spec_path 2026-02-20 13:46:59 +05:30
Sameer Kankute
b579f56ca3 Fix mapping of parallel_tool_calls for bedrock converse 2026-02-20 12:27:41 +05:30
Sameer Kankute
cdb4917185 Fix: api_base is required. Unable to determine the correct api_base for the request 2026-02-20 12:09:25 +05:30
yuneng-jiang
527f5c5d14
Merge pull request #21657 from BerriAI/litellm_knip_unused_files_cleanup
[Refactor] UI: Remove 38 unused files detected by knip
2026-02-19 22:22:21 -08:00
yuneng-jiang
eeab8705a1 [Refactor] UI: Remove 38 unused files detected by knip
Dead code cleanup — these files had no imports from any active entry points.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-02-19 22:07:41 -08:00
Sameer Kankute
5885c4a588 Fix _map_reasoning_effort_to_thinking_level for all gemini 3 family 2026-02-20 11:17:24 +05:30
yuneng-jiang
6bfab8acd4 address greptile review feedback (greploop iteration 2)
Reset logo_path to default_logo when custom UI_LOGO_PATH file doesn't
exist, so the else branch at the bottom of get_image serves the default
logo instead of the non-existent custom path.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-02-19 20:25:00 -08:00
jquinter
37c98f8325
Merge pull request #21602 from BerriAI/fix/policy-endpoints-unused-import
fix(lint): remove redundant router import in policy_endpoints __init__
2026-02-20 01:16:18 -03:00
jquinter
d08c3a1d68
Merge pull request #21638 from BerriAI/fix/ai-policy-suggester-use-litellm-acompletion
fix(policy): use litellm.acompletion directly in AiPolicySuggester
2026-02-20 01:14:59 -03:00
jquinter
898774fecf
Merge pull request #21643 from BerriAI/fix/redis-cache-test-missing-host
fix(tests): pass host to RedisCache in test_team_update_redis
2026-02-20 01:11:04 -03:00
yuneng-jiang
145efe2267 address greptile review feedback (greploop iteration 1)
Add os.path.exists check before serving custom local logo so that a
non-existent UI_LOGO_PATH gracefully falls through to the cache/default
instead of causing a FileResponse error.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-02-19 20:09:42 -08:00
Sameer Kankute
fb75a7130f
Merge pull request #21630 from Chesars/fix/empty-system-message-anthropic
fix(anthropic): empty system messages in translate_system_message
2026-02-20 09:32:50 +05:30
Julio Quinteros Pro
bd8c1cc673 fix(tests): pass host to RedisCache in test_team_update_redis to avoid ValueError
RedisCache() without arguments fails at construction with
"ValueError: Either 'host' or 'url' must be specified for redis."
The actual Redis connection is irrelevant since async_set_cache is mocked.
Unlike test_get_team_redis which uses client_no_auth (which sets REDIS_HOST
via fake_env_vars), test_team_update_redis has no fixture setting that env var.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-02-20 01:00:58 -03:00
jquinter
6b681aafdd
Merge pull request #21641 from BerriAI/fix/jwt-tests-missing-premium-user-flag
fix(tests): set premium_user=True in JWT tests that call user_api_key_auth
2026-02-20 00:45:13 -03:00
Julio Quinteros Pro
cde51a53a7 fix(tests): set premium_user=True in JWT tests that call user_api_key_auth
JWT auth is an enterprise-only feature. Tests that call user_api_key_auth
with enable_jwt_auth=True must set premium_user=True on the proxy server
to bypass the enterprise gate, otherwise they fail with:

  ValueError: JWT Auth is an enterprise only feature.

This follows the same pattern as PR #21285 (fix/jwt-enterprise-license-test).

Fixed tests:
- test_team_token_output
- test_allowed_routes_admin
- test_allow_access_by_email
- test_end_user_jwt_auth

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-02-20 00:31:56 -03:00
jquinter
2b09c5d3bc
Merge pull request #21639 from BerriAI/fix/unused-imports-semantic-guard-and-policy-endpoints
fix(lint): remove unused imports in semantic_guard and policy_endpoints
2026-02-20 00:02:21 -03:00
Sameer Kankute
b405f64688
Merge pull request #21598 from jtsaw/fix/anthropic-sonnet-4-6-reasoning-effort
support reasoning and effort parameters on sonnet 4.6
2026-02-20 08:24:29 +05:30
Julio Quinteros Pro
dba60cac22 fix(lint): remove unused imports in semantic_guard and policy_endpoints
- Remove unused RouteChoice import from semantic_guard.py TYPE_CHECKING block
  (only referenced in docstring, never used as a type annotation)
- Remove redundant explicit router import from policy_endpoints/__init__.py
  (already re-exported by the preceding import * on the same line)

Fixes Ruff F401 errors caught in CI linting.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-02-19 23:46:02 -03:00
Julio Quinteros Pro
3381d6217f fix(policy): use litellm.acompletion directly in AiPolicySuggester
The suggest() method was calling llm_router.acompletion(), which requires
the proxy router to be initialised. This caused:
  ValueError: LLM router not initialized
in every test (and in any deployment without a configured router).

AiPolicySuggester is a self-contained feature that calls an LLM directly;
it has no need for proxy routing. Switch to litellm.acompletion, which is
always available and is what the tests already mock.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-02-19 23:35:35 -03:00
yuneng-jiang
a8026154ab [Fix] /get_image returns stale cached logo instead of custom UI_LOGO_PATH
The /get_image endpoint checked for cached_logo.jpg before reading the
UI_LOGO_PATH env var, so a pre-existing cache (e.g. baked into the base
Docker image) would always be served, ignoring the user's custom logo.

Move the UI_LOGO_PATH read before the cache check and serve local file
paths directly, bypassing the cache. The cache optimization is preserved
for HTTP URLs and the default logo where it is actually needed.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-02-19 18:32:48 -08:00
Ryan Crabbe
dc13378505 feat: add AI-powered duplicate PR detection via opencode
Same approach as the duplicate issue detector — uses opencode run
with gh pr commands to find potentially duplicate open PRs when
external contributors open new PRs. Skips core team and bots.
2026-02-19 17:51:44 -08:00
Darien Kindlund
0f2516e9cd
fix: handle explicit None model_info in LowestLatencyLoggingHandler (#21633)
When model_info is explicitly set to None in litellm_params (via
get_litellm_params.py), the pattern .get("model_info", {}) returns
None instead of {}, causing AttributeError on the chained .get("id").

This commonly occurs when Anthropic API returns usage limit errors,
the router falls back to Vertex AI, and the fallback succeeds — the
log_success_event callback fires with model_info=None.

Uses the (x or {}) pattern already established in router.py.

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-02-19 17:30:58 -08:00
Ishaan Jaffer
d2bdf7d618 fix form 2026-02-19 17:17:20 -08:00
Ishaan Jaffer
be16f320ac fix filter 2026-02-19 17:17:11 -08:00
Ishaan Jaffer
2199b0783d ui refactor 2026-02-19 17:13:51 -08:00
Ishaan Jaffer
35120515cb QA: UI fix test page 2026-02-19 17:13:51 -08:00
yuneng-jiang
aa9f85545c
Merge pull request #21627 from BerriAI/litellm_service_key_vis
[Fix] Service Account Visibility for Team Members
2026-02-19 17:09:33 -08:00
Ryan Crabbe
974392311d perf: short-circuit is_model_o_series_model with startswith before set lookup
Reorder the check to use str.startswith(tuple) first, which immediately
returns False for non-o-series models (the common case), avoiding the
genexpr + 198-element set lookup. Line profiling shows 4.75x speedup
(13.6µs → 2.9µs per call, 2.45s → 0.52s across 180k calls).
2026-02-19 16:58:47 -08:00
yuneng-jiang
4475258426 address greptile review feedback (greploop iteration 1)
Refactor to fetch team objects once via _fetch_user_team_objects(),
then derive admin and member team IDs from the shared result. Eliminates
duplicate DB query between get_admin_team_ids and get_member_team_ids.
2026-02-19 16:53:48 -08:00
Chesars
56386969b5 fix(anthropic): remove empty system messages from message list
Empty system messages were skipped for Anthropic's system param but
not removed from the messages list, causing BadRequestError when
anthropic_messages_pt encountered the unsupported "system" role.

Fixes #21622
2026-02-19 21:46:17 -03:00
Chesars
3564b8d83b fix(types): suppress Pydantic serialization warnings on ModelResponse choices
Pydantic v2's Union serializer for `List[Union[Choices, StreamingChoices]]`
tries both branches when serializing, emitting spurious
`PydanticSerializationUnexpectedValue` warnings (field count mismatch on
`Message` and type mismatch `Expected StreamingChoices but got Choices`).

Add a `WrapSerializer` on the `choices` field that serializes each item
individually via its own `model_dump()`, bypassing the Union dispatch
entirely while correctly propagating `exclude_none`, `exclude_unset`,
and `exclude_defaults` from the parent serialization context.
2026-02-19 21:43:19 -03:00
yuneng-jiang
c7fe400d4d [Fix] Service account visibility for team members
Regular team members could not see service accounts belonging to their
team. Additionally, former team members could still see service accounts
they created via the created_by filter after leaving the team.

- Add get_member_team_ids() to retrieve all teams a user belongs to
- Regular members now see team service accounts (user_id=NULL) but not
  other members' personal keys
- Scope created_by filter to the user's current team memberships
- Add 7 unit tests covering the new visibility rules
2026-02-19 16:29:43 -08:00