mirror of
https://github.com/BerriAI/litellm.git
synced 2026-08-28 05:25:59 +00:00
1187 commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
ryan-crabbe-berri
|
7d5a2c1a0d |
Merge remote-tracking branch 'origin/litellm_internal_staging' into litellm_ruff_dead_test_code
# Conflicts: # ruff-tests.toml |
||
|
yuneng-jiang
|
6a0d03914c
|
test: drop the cwd-relative sys.path.insert calls from the test suite (#37802)
* test: drop the cwd-relative sys.path.insert calls from the test suite
TQ003 stands at 1,077 across 1,058 files, and 1,015 of them are the same shape:
sys.path.insert(0, os.path.abspath("../..")) and its deeper siblings. The
argument resolves against the working directory rather than the file, so from
the repo root, where every job runs pytest, it inserts the directory two levels
above the checkout. It has never pointed at litellm. The package is installed
into the environment anyway, which is what actually makes the import work, and
what the rule's message has said all along.
Removing them leaves 1,634 imports of sys and os with no remaining reference,
and those go too, except where another test module imports the name back out of
the file. The rest of TQ003 is 62 call sites that resolve against __file__ or a
variable, which are a different question and are left alone.
Collection is identical either way: 45,871 tests and the same 51 pre-existing
collection errors before and after, and ruff reports no new undefined name.
* test: drop the duplicate imports the sys.path sweep exposed to F811
* test(pre-call-utils): restore the os import the new bedrock tests need
|
||
|
yuneng-jiang
|
de1bc29dc7
|
test: unshadow the module handles the F811 sweep left behind (#37914)
* test: unshadow the module handles the F811 sweep left behind, and pin the two live tests that went red with it The F811 sweep in #37878 removed the fixture-local `import litellm` from four conftests, but the bare `import litellm.proxy.proxy_server` a few lines below still binds `litellm` as a function local, so `importlib.reload(litellm)` runs before the name is assigned and every test in those directories errors at setup. The `hasattr` guard on the line above already proves the module is loaded, so the import only ever bound the name. Drop it, and enable F823 in ruff-tests.toml, which flags all four sites at the failing line and would have blocked the sweep The same sweep renamed the `check_non_streaming_response` parameter but left one read of `completion`, which now resolves to `litellm.completion`, and removed an import whose side effect was the only thing making `litellm.proxy.proxy_server` reachable in the moderation hook test. That test already takes `monkeypatch`, so patch the router through it and stop leaking the router into later tests `test_content_policy_exception_openai` passed vacuously until #37887 turned it into a real `pytest.raises`, and OpenAI no longer rejects a lyrics prompt with a content policy error. Inject an AsyncOpenAI client whose transport answers with OpenAI's own `content_policy_violation` rejection so the mapping to ContentPolicyViolationError is exercised every run `test_async_create_batch` hit a 409 cancelling a batch OpenAI had already marked failed. The cancel step tolerated a completed batch but not a failed one. Fold both guards into one helper that tolerates a failed batch only when OpenAI's recorded error is the org's enqueued token limit, and prints the batch's errors so the reason is in the log either way * test: close the injected AsyncOpenAI client after the content policy test * chore(lint): ratchet TQ005 down by the global mutation this branch cleared * chore(lint): ratchet TQ005 to 2660 on the merged tree * chore(lint): ratchet TQ005 to 2561 on the merged tree * chore(lint): ratchet TQ005 to 2548 on the merged tree |
||
|
ryan-crabbe-berri
|
b7f8016002 |
test: gate the test suite on F601, B023, B025 and F632
Four more ruff rules for code the test suite runs but never checks. F601 is the one that paid: the duplicate key it flagged in a get_form_data fixture was the mock reproducing the production bug fixed in the previous commit. B025 removed two unreachable handlers, one of them a pytest.skip shadowed by an earlier `pass`, so an upstream Vertex flake reported green having asserted nothing. F632 turned an `is ""` identity check, which passes only on CPython interning, into the `== ""` it meant. B023 fixed three closures over loop variables, all latent today but one iteration-order change away from checking the last case N times. |
||
|
ryan-crabbe-berri
|
6b088f4bb1 | style: wrap the escaped messages under 120 columns | ||
|
ryan-crabbe-berri
|
91599aef69 | test: say whether a match= pattern is a regex or a literal (ruff RUF043) | ||
|
ryan-crabbe-berri
|
5ed230701a | test: escape the literal match= patterns PT017 minted | ||
|
ryan-crabbe-berri
|
4d8346a5b9 | test: wrap the raising call, not the print that follows it | ||
|
ryan-crabbe-berri
|
243ed4393d |
test: reject assertions on a caught error inside except (ruff PT017)
A test that asserts on the error inside its own except block passes when the call stops raising, because nothing runs the handler. That is the exact case the test exists to catch, so the regression lands green. Rewrites all 111 such blocks into pytest.raises, which fails when the call succeeds, and selects PT017 in ruff-tests.toml so no new one lands. |
||
|
ryan-crabbe-berri
|
e9d40a8f73 |
test: enforce F811 so a duplicate definition cannot silently replace the first
A name bound twice keeps only the second binding. In `tests/` that is nearly always a repeated import, harmless but misleading, and the same rule is what catches the cases that are not harmless: a local that shadows an import the module still calls, and a second `def test_x` that quietly replaces the first. 311 of the 344 sites were repeated imports and came out with ruff's own fix. The remaining 33 needed a decision. Four modules imported a name they never used because a local definition below already shadowed it. Two comprehensions bound `call` over `unittest.mock.call`, which those modules import and use. One test rebound the two module handles its nested reload closure had captured. One class attribute shadowed an unused `status` import. The load-test fixtures move to a conftest, which is how pytest is meant to share them, so the test module no longer imports three fixture names it never calls. The nine `prisma_client` parameters keep a narrow `noqa`: pytest resolves that fixture by name before the body runs, so the parameter never shadows anything. |
||
|
ryan-crabbe-berri
|
b76def0e5d
|
test: require a match= on broad pytest.raises, and drop duplicate parametrize cases (#37769)
`pytest.raises(Exception)` with no `match=` passes on any error that broad. A TypeError from a refactor, a botched fixture, an import that moved: all of them read as the rejection the test claims to police, so the test goes green for the wrong reason and stays green after the behaviour it guards is gone. PT011 closes that gap for the 317 sites B017 could not reach, because B017 only fires on a single-statement body with no `as e` binding. Each pattern here is the message the code actually raised, recorded by running the sites under a plugin that logged the concrete type and text per call site, so the assertions describe observed behaviour rather than a guess. Where a site raises more than one message across its parametrize cases, the pattern is an alternation of what was seen; where the exception carries an empty `str()` and puts the text on `.message`, the site keeps a narrow `noqa` with the reason. PT014 removes four parametrize cases that were listed twice. The duplicate re-runs an assertion that already passed, and it usually marks a case someone meant to vary and forgot to edit. |
||
|
ryan-crabbe-berri
|
a112ba5f63
|
test: enforce PT012 so a pytest.raises block cannot hide dead assertions (#37748)
* test: enforce PT012 so a pytest.raises block cannot hide dead assertions `with pytest.raises(...)` stops at the first statement that raises. Anything sequenced after it inside the block never runs, so an assertion written there is never checked and the test still reports green. Two sites were doing exactly that, and both assertions turned out to be wrong once they started running. tests/llm_translation/test_prompt_factory.py asserted the bedrock rejection names "requires at least one non-system message", which holds. tests/proxy_unit_tests/test_proxy_server.py asserted the prisma startup failure mentions "httpx.ConnectError", which never appears: the failure is an httpx.ConnectError whose message is "All connection attempts failed", so that test now asserts the type. Its DATABASE_URL override moves to monkeypatch, since the old restore sat below the assertion and leaked the invalid URL into every later DB test the moment the assertion started being able to fail. The remaining 72 sites are rewritten without changing what they exercise: setup that cannot raise moves above the block, a nested `patch` moves outside it, and bodies with real control flow (a stream drain, an if/else on sync_mode, a retry loop) move into a local closure the block calls. Fixing PT012 unmasked two B017s, since ruff only reports a blind pytest.raises(Exception) once the block holds a single statement. tests/proxy_unit_tests/test_auth_checks.py narrows to the ProxyException can_key_call_model actually raises. tests/local_testing/test_completion_cost.py was asserting vertex_ai/medlm-medium has no cost entry, which stopped being true at some point; that dead first half is gone and the rest of the test, which checks medlm pricing resolves above zero, now runs instead of being skipped. * chore(ci): ratchet TQ004 to 768 after the prisma test moved to monkeypatch |
||
|
yuneng-jiang
|
35416c702d
|
test: point the live together_ai suites at a model together still serves (#37746)
Every live together_ai call in CI has answered 503 Service unavailable since 2026-08-20, across two runs 2.5 hours apart, while Together's status page reported no incident in either window. These are real calls, not replayed cassettes: the VCR layer runs filter_non_2xx_response, so a 503 is never written to a cassette and cannot be replayed back. Qwen/Qwen2.5-7B-Instruct-Turbo does not appear anywhere on Together's monitored component list, whose Qwen entries are all Qwen3.x, so a model-level outage there would never surface as an incident. The same 503 already forced test_basic_rerank_together_ai to be skipped on a different together_ai model, so per-model 503s are an established failure mode here rather than a platform outage. openai/gpt-oss-20b is the cheapest together_ai entry that carries real pricing and the capabilities these suites exercise, at $0.05/$0.20 per 1M tokens with function calling, response schema and tool choice. Together monitors it as a served component. The retired model also carries null pricing in the cost map, which is its own liability now that unpriced models are blocked. test_multiple_deployments.py keeps the old id: it is a router fallback list that is green today, and busting its cassette to prove a point would trade a passing test for a live call this change cannot vouch for. |
||
|
ryan-crabbe-berri
|
680bcfd8aa
|
test(lint): ban blind pytest.raises(Exception) with ruff B017 (#37731)
* test(lint): ban blind pytest.raises(Exception) with ruff B017 A bare pytest.raises(Exception) accepts whatever the body throws. The TypeError a refactor introduces satisfies it exactly as well as the rejection the test was written for, so the crash reads as a pass and the test never goes red. All 111 existing sites are narrowed here. A runtime probe recorded the concrete exception each one actually catches, and each site now names that type. Where the code under test genuinely raises a bare Exception, the site pins a stable slice of the message with match= instead. Two sites tell on themselves. The shared responses-API cancel test raises "custom_llm_provider is required but passed as None" rather than talking to a provider at all, because cancel_responses takes a provider, not a model. And test_bedrock_guardrails_with_streaming was the only test in its file still passing without AWS credentials, because the NoCredentialsError boto3 raised long before the guardrail ran satisfied the blind raises. * fix(test): widen the openai batch-dispatch assertion to OpenAIError The narrowed NotFoundError only holds where OPENAI_API_KEY is set. Without one the SDK raises OpenAIError while building the client, long before any 404, so CI went red. OpenAIError covers both and still rejects a TypeError from a refactor. |
||
|
yuneng-jiang
|
0c2e404be3
|
test(ci): serve /moderations from the canned OpenAI mock (#37739)
* test(ci): serve /moderations from the canned OpenAI mock The otel proxy E2E job points its `openai/*` wildcard deployment at the canned mock, and #37492 made `get_model_list` agree with `get_available_deployment` on bare model names. /moderations now resolves `omni-moderation-latest` to that wildcard deployment the way /chat/completions already did, so the request lands on the mock, which never implemented the route and answers a bare 404. Add /moderations and /v1/moderations to the mock, returning an OpenAI-shaped response with one result per input item. * style(ci): annotate the new moderations locals as Final |
||
|
ryan-crabbe-berri
|
21e9632713
|
test: add six ruff rules that catch tests which cannot fail (#37709)
`assert False` inside a `try:` raises AssertionError, which the `except Exception` right below it catches, so several tests reported green no matter what the code did. `pytest.fail` raises Failed, a BaseException, and escapes. A bare `a == b` statement is evaluated and discarded. Nine of those sat in tests, and one was comparing against a model name the router never produces. Selects B011, B015, B018, PT015, PLR0133 and PLW0127 in ruff-tests.toml alongside F821, with all 50 existing violations fixed, so no budget file or ratchet is needed. CI already runs this config over tests/. |
||
|
ryan-crabbe-berri
|
4af59d7c6e
|
ci: lint the test tree for undefined names and fix all 30 (#37671)
ruff.toml excludes tests/* from `ruff check`, so nothing has ever checked the test tree for names that do not exist. That matters more in tests than in product code: a NameError inside a test whose body is wrapped in `except Exception: pass` is swallowed, and the test reports green forever. Adds ruff-tests.toml selecting F821 alone, wired into the lint workflow and `make lint-ruff`, and clears every existing violation: - 4 tests interpolated an unbound `e` into a `pytest.fail` message reached only on the failure path, so the NameError, not the assertion, is what ran. test_llm_guard_error_raising is the worst: it passes today with content safety disabled entirely. It now asserts the 400 and its detail body. - 5 sites construct BaseExceptionGroup, a 3.11 builtin, in a tree that still supports 3.10. Guarded behind the exceptiongroup backport that anyio already pulls in below 3.11. - 9 missing imports (json, openai, Any, Final, HTTPException), including one in a helper that catches HTTPException by a name it never imported, so the challenge path it exists to detect raises NameError instead. - 5 annotations naming types imported inside the function body, hoisted to module scope or TYPE_CHECKING. - 2 blocks of dead code: everything after a pytest.fail in test_claude_agent_sdk, and an unused helper in test_end_users calling a function defined in a different module. - 1 error-path f-string in the router-settings doc test that masked the real FileNotFoundError behind a NameError. Only F821 for now. Widening the select list means ratcheting thousands of pre-existing findings, so rules go in one at a time with their violations already fixed. |
||
|
ryan-crabbe-berri
|
487356733c
|
test: replace blind sleeps with deadline waits in callback and caching tests (#37660)
* test: replace blind sleeps with deadline waits in callback and caching tests tests/local_testing/test_custom_callback_input.py slept a fixed 1-3s after every call and then asserted the callback handler recorded no errors. Because the handler only appends to `states` when a callback actually fires, an assert of `len(errors) == 0` passes just as happily when nothing fired at all, so the sleep was buying flakiness in exchange for a vacuous check. The async tests were worse: `time.sleep` blocks the event loop, so the success/failure tasks scheduled on it could not run before the assertion. Adds tests/_wait_helpers.py with `wait_until` / `await_until`, which poll a predicate against a deadline, and converts all 17 sites to wait on the thing the test actually cares about (the terminal state landing in `states`, or the patched log hook being called). The waits assert the callback fired, so these tests now fail on a dropped callback instead of passing silently. The three sleeps in test_caching_handler.py sat between `sync_set_cache` and `_sync_get_cache`, both fully synchronous against a local in-memory cache, so they are just deleted. * fix(test): wait on the priming call's own logging in the cache-hit test The 3s sleep in test_logging_async_cache_hit_sync_call was not waiting for the cache write, which lands before the stream iterator is exhausted. It was waiting for the priming call's success callback to drain, so the handler installed right after it only ever sees the second, cache-hit call. Waiting on a populated cache_dict let the priming call's still-pending log_success_event reach the new mock, and the test then read cache_hit off the wrong payload. Waits on the priming handler's own sync_success state instead. |
||
|
yuneng-jiang
|
76aa13cde0
|
test: remove the five test functions a later definition shadows (#37591)
Python binds a name once per scope, so when a module or class defines the same test twice only the last one exists. The earlier definitions are unreachable: pytest never collects them, and nothing that references them can fail. A sweep in August cleared nine of these. Five have appeared since, which is the argument for a rule rather than another sweep. Each survivor is the better version, so nothing is lost. The two SQS logger twins additionally stub `asyncio.create_task`, which the shadowed copies did not. The cost-calculator duplicate is a two-line stub that also takes a `model_item` parameter no fixture supplies, so it could not have run even unshadowed. The two `test_prompt_caching` bodies are both `pass`. Collecting the four files reports 416 tests before and after. `tests/proxy_unit_tests/conftest copy.py` goes with them. pytest only loads a file named exactly `conftest.py`, nothing imports this one, and the space in the name says what it was. |
||
|
yuneng-jiang
|
bcead282e2
|
test: move the remaining live groq call sites off the retired llama models (#37426)
The earlier sweep only caught the conformance suite in tests/llm_translation. Groq retired llama-3.1-8b-instant alongside llama-3.3-70b-versatile, and four tests under tests/local_testing still call them for real, so litellm_router_testing and both local_testing shards 404 with model_not_found. Only the sites that leave the process move. The chunk fixtures in test_stream_chunk_builder, and the cost and routing tests that never open a socket, keep the old ids because the string is data there, not a request. |
||
|
tin-berri
|
1f4acbb924
|
feat(complexity_router): custom classifier plugins via classifier_type 'custom' (#37249)
* feat(complexity_router): custom classifier plugins via classifier_type 'plugin' Adds a third classification mode where an operator-supplied hook decides the tier instead of the heuristic scorer or the LLM classifier. The hook implements an async classify(context) returning a tier name (built-in value, tier_labels label, or tier_definitions name) or None to decline; failures, timeouts, and unknown tiers fall back exactly like a failed LLM classifier. The context carries the request messages and metadata, including caller identity, so a plugin can route by team, spend, or any business rule. The plugin resolves from a dotted path at proxy startup with a load-time check that classify is a coroutine function, and is closed off over HTTP like the routing plugins list. Routing decisions record the new classifier_plugin cause. tier_definitions now accepts classifier_type 'plugin' alongside 'llm'. * fix(proxy): resolve plugin dotted paths in _delete_deployment before hashing ids The db-sync reconcile re-reads the raw config and hashes litellm_params to compute which ids the config wants served, but the router's ids were hashed from the resolved params where plugin dotted paths are live instances. The mismatched ids made the reconcile evict every plugin-bearing auto-router one sync after startup, on any proxy with a database connected. This also affected the existing routing plugins list, not just the new classifier plugin. Resolving the plugins in _delete_deployment the same way load_config does makes both sides hash the same canonical form. A plugin module broken on disk at reconcile time skips cleanup instead of evicting valid deployments, matching how a get_config failure is handled * fix(complexity_router): treat non-string plugin verdicts as declines, centralize the empty-mapping sentinel A hook returning a non-string raised inside resolve_classified_tier outside the plugin exception boundary, failing the request instead of falling back. Also moves the read-only empty mapping to constants.py per repo convention and moves the classifier plugin product docs out of the package README for the docs repo * refactor(complexity_router): rename the plugin classifier mode to classifier_type 'custom' The mode value now names the operator's intent while classifier_plugin keeps naming the mechanism; routing decisions keep the classifier_plugin cause * refactor(proxy): pin plugin-bearing deployment ids from the raw params instead of resolving in the reconcile Replaces the previous approach of re-running plugin resolution inside _delete_deployment, which imported operator modules on every reconcile cycle and skipped the whole cleanup pass when any one module was broken on disk. load_config now stamps model_info.id from the raw litellm_params before resolution swaps dotted paths for live instances, so the reconcile's raw-config hash matches by construction and needs no resolution at all: a broken module cannot stall cleanup for unrelated models, and any future param-transforming resolution is covered by the same pin. _generate_model_id becomes a staticmethod so the pin can run before the Router exists; its statically dead non-string key branches are removed. Also documents candidate_models as an informational snapshot for classifier plugins, unlike the narrowing surface RoutingPlugin filters * fix(router): restore _generate_model_id key handling, align classifier context with the routing-plugin pattern The staticmethod conversion accidentally dropped the non-string-key branches from _generate_model_id, a silent hash change for any params with non-string keys; they are restored verbatim. The classifier plugin context now follows the Router-level routing-plugin recipe exactly: structured messages come from resolve_structured_messages over the raw messages, and the metadata key comes from the shared get_metadata_variable_name_from_kwargs helper, which also replaces the duplicated inline sniff in _pick_model_for_tier. This removes the raw-or-resolved fallback where a plugin could silently receive resolved messages when a call site forgot to pass the raw ones * refactor(router): make generate_model_id public, guard classifier context construction Two modules legitimately hash deployment ids with the same helper now (Router and the proxy's config-load pin), so the private name was lying about its audience and the cross-module call needed a pyright suppression; renaming it public restores the static safety net. The classifier plugin's RoutingContext construction moves inside the failure boundary, matching the LLM path where litellm-side prompt building also falls back rather than failing the request, and a prompt-only call with no message list is now covered by a test |
||
|
Yuneng Jiang
|
075781568d
|
test: remove tests that never execute
Three groups, all verified by running the suite rather than by inspection. 18 files whose every test function carries an unconditional @pytest.mark.skip, 39 test functions in total. They are collected on every CI run and always skip, so they advertise coverage the suite does not have. Reasons on the marks include "AWS Suspended Account", "lakera deprecated their v1 endpoint" and "moved to using 'otel' for logging"; 26 of the marks predate 2025. 30 test functions with a byte-identical body and identical decorators to a sibling in the same file and class, differing only in name. Deleting one of each pair removes no coverage. Four further candidates were excluded because they override an inherited test, where deleting the override un-shadows the base class implementation instead of removing a duplicate. 9 test functions that a later definition of the same name shadows, so Python never binds them and pytest cannot collect them. One file that is a demo script rather than a test; its own docstring says to run it with python. Verification: collecting the 26 edited files gives 2,492 node IDs before and 2,462 after. The 30 duplicate deletions account for exactly 30 removals, the 9 shadowed deletions account for 0 (confirming at runtime that they were never collectable), nothing unexplained disappeared, and nothing new appeared. No other test or module imports any deleted symbol. |
||
|
yuneng-jiang
|
5e620af405
|
Merge pull request #36600 from BerriAI/litellm_/bedrock-retired-sonnet-test-model
test(bedrock): repoint live Claude tests off the retired Claude 3 Sonnet |
||
|
Yuneng Jiang
|
d49114b101
|
test(bedrock): repoint live Claude tests off the retired Claude 3 Sonnet
AWS no longer serves `anthropic.claude-3-sonnet-20240229-v1:0`. The streaming path returns a plain 404, "Model with the provided id anthropic.claude-3-sonnet-20240229-v1:0 is not found", and the non-streaming path answers 500 for the same reason. Our own cost map has carried a 2026-07-30 deprecation date for it since #36538 That accounts for 20 failures across local_testing_part1, local_testing_part2 and llm_translation_testing. litellm maps both statuses correctly, so the tests are what went stale, not the client Replacement is `us.anthropic.claude-sonnet-4-5-20250929-v1:0`: a like-for-like Sonnet, and the newest Bedrock Sonnet this repo exercises against the real API in tests/e2e. Newer ids exist in the cost map, but nothing in the repo calls them live, so picking one would be an unverified guess about model access on the CI account Scope is limited to the tests that actually issue a request. The occurrences that assert on the model string itself, or that feed mocked transformations, keep the old id so their assertions stay meaningful |
||
|
Yuneng Jiang
|
5669742ea6
|
fix(model_prices): advertise native structured output on every Bedrock DeepSeek V3.2 and GLM 5 id
`supports_native_structured_output` was set only on the bare `deepseek.v3.2` and `zai.glm-5` entries, so the cross-region inference profiles and the region-pinned ids resolved to None. The flag gates the native `outputConfig.textFormat` branch in BedrockConverseConfig, so callers addressing the same model as `us.deepseek.v3.2` or `bedrock/us-west-2/deepseek.v3.2` silently fell back to synthetic tool injection. `us.` is the form Bedrock steers callers toward, so the most common way to reach these models was the one missing the capability. Adds the flag to the 12 affected ids and keeps the packaged backup in sync. test_get_model_info_bedrock_models already caught the region-pinned ids, but it filters on `litellm_provider == "bedrock"` and the cross-region profiles carry `bedrock_converse`, so reverting just `us.deepseek.v3.2` and `eu.deepseek.v3.2` left it green. The new parity test covers the prefixed profiles and fails on exactly that mutation. |
||
|
Yuneng Jiang
|
b6557d2b14
|
test: repair three failing suites on litellm_internal_staging
The management route-coverage guard fires because /team/metadata_schema landed in #33353 without a behavior-suite scenario, so this adds one covering the nine seeded actors plus the unauthenticated 401 The prometheus budget-metric assertions read the log call's first positional arg, which #35703 turned into an unrendered "%s" format string when it moved logging to lazy args. They now render the message from the call args, which also pins the arg order and the exception text that the old substring check never reached GitHub Models was fully retired on 2026-07-30, so test_completion_github_api can no longer pass: the endpoint the github provider targets returns 404 and models.github.ai answers 410 "github_models_retirement_brownout". The dead live test is removed rather than skipped |
||
|
Yuneng Jiang
|
67dd8924ee
|
test(proxy): assert _delete_deployment's still-desired id set instead of a delete count
_delete_deployment stopped returning a count of evictions in #35400 and now returns the frozenset of ids the db and config still want, so a caller judging its own reload can tell a deliberate eviction from a deployment that went missing. These two tests in tests/local_testing were left comparing that frozenset against an int and have been failing since; the directory is only referenced by .circleci/config.yml, which no longer reports checks on PRs, so nothing caught them. The eviction behavior itself is unchanged, so the fix is on the assertions: compare against the expected id set, and pin the router's surviving ids so a mutation that evicts the wrong deployment is caught rather than passing a bare length check. |
||
|
Yuneng Jiang
|
ffd6ac52c5
|
fix(deps): raise aiohttp floor to 3.14.2 to clear pooled-connection timeouts
aiohttp 3.14.0 and 3.14.1 re-arm the sock_read timer on a keep-alive connection after it has already been returned to the idle pool. The stray timer stamps a SocketTimeoutError on the pooled connection without closing it, so the pool keeps handing it out and the next request to pick it up fails instantly on an error left behind by an earlier, unrelated request. Because a single pool is shared across providers, the failures appear simultaneously across Vertex AI, Bedrock, Anthropic and OpenAI-compatible deployments as sub-millisecond "Connection timed out" errors. uv.lock resolved aiohttp 3.14.1 and the published images install via `uv sync --frozen`, so every image built from that lock shipped the regression. The wheel's own metadata declared `aiohttp>=3.10,<4.0`, which also left pip consumers free to resolve into the same broken window, so both the runtime floor and the uv constraint move to >=3.14.2. Upstream fixed this in aio-libs/aiohttp#12954, released in aiohttp 3.14.2; the lock now resolves 3.14.3. Raising the floor rather than capping below 3.14 keeps the advisories that the existing 3.14.1 floor cleared, so no osv-scanner ignores are needed. litellm requires Python >=3.10 and aiohttp 3.14.2 requires >=3.10, so no supported interpreter loses support. Both new tests fail on the previous pins and pass on these. |
||
|
devin-ai-integration[bot]
|
e59add11cd
|
fix(anthropic): self-heal on missing thinking-signature errors from Bedrock/Vertex (#33719)
* fix(anthropic): self-heal on missing thinking-signature errors from Bedrock/Vertex Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * fix(anthropic): narrow thinking signature error marker Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * test(router): stabilize prompt caching fixture size Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * chore: re-trigger CI Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> --------- Co-authored-by: Krrish Dholakia <krrishdholakia@berri.ai> Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> |
||
|
devin-ai-integration[bot]
|
44d9737609
|
fix(llm_guard): apply sanitized prompt returned by moderation API to request (#33331)
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> |
||
|
mateo-berri
|
77885779ca | refactor(fallback-generalizations): split rules into routing and provider-neutral capability kinds | ||
|
Mateo Wang
|
684e3e1c2e
|
test(vertex_ai): bump local_testing vertex tests from gemini-2.5-flash to gemini-3.5-flash (#32439) | ||
|
Mateo Wang
|
b2e2a38bc0
|
fix(passthrough): stream non-sse passthrough responses instead of buffering in memory (#32386)
* fix(passthrough): stream non-sse passthrough responses instead of buffering in memory Non-SSE passthrough responses were fully read into proxy memory (content = await response.aread()) before the first byte reached the client. For large non-JSON bodies such as Anthropic batch results jsonl files this ballooned proxy RSS to a multiple of the file size and produced near-total TTFB dead air, letting intermediaries kill the silent connection and truncate the download. The upstream request is now sent with httpx stream semantics and the buffering decision is made from the response headers: application/json (and +json) bodies plus upstream errors keep the buffered behavior since spend logging, guardrails and managed-id rewriting inspect them, while every other 2xx body is relayed as a StreamingResponse that iterates upstream bytes without accumulating them, preserving status code and headers (including x-litellm-*) and firing the success-handler logging with response_body=None once the stream completes. * fix(passthrough): log client disconnects mid-stream and derive test client cache key from production code * test(passthrough): intercept AsyncClient.send in legacy passthrough tests and assert final wire params * test(passthrough): fail with a clear assert when the passthrough client cache scan misses |
||
|
Mateo Wang
|
d0c82c308d
|
fix(main): stop per-request custom pricing from clobbering shared model_cost pricing (#32163)
* fix(main): stop per-request custom pricing from clobbering shared model_cost pricing
A request routed through a wildcard deployment with explicit zero pricing
(e.g. openai/* with input_cost_per_token: 0) registered that pricing on the
shared {provider}/{model} key in litellm.model_cost, so sibling deployments
relying on built-in pricing logged $0 until process restart (LIT-3991).
Request-time registration in completion()/embedding() now mirrors the
router-startup isolation: router-originated requests register full pricing
under the deployment's unique model id only, while the shared backend key
receives the entry with custom pricing fields stripped. Direct SDK calls
without a router deployment id keep the legacy shared-key registration.
The stripping logic is shared via
CustomPricingLiteLLMParams.strip_custom_pricing_fields and reused by
Router._create_deployment and Router.add_deployment.
* test: update legacy tests that asserted per-request pricing leaking into shared model_cost
test_router_fallbacks_with_custom_model_costs asserted the shared
claude-sonnet-4-5-20250929 entry ends up with the deployment's 30/60
pricing, which is exactly the cross-deployment leak this PR removes; it
now asserts the shared key keeps the built-in pricing, matching the
test's stated goal.
test_cost_calc.py::test_run computed streaming cost via
completion_cost(response), which only matched the non-stream cost while
the shared gpt-3.5-turbo entry was poisoned with the per-request
2/token pricing; it now passes the request's custom pricing explicitly
via custom_cost_per_token.
|
||
|
Mateo Wang
|
ee3debe82e
|
fix(dynamic_rate_limiter): inject clock so active-project window is stable within a request (#32299) | ||
|
Sameer Kankute
|
64dc5080b9
|
fix(bedrock): drop strict/additionalProperties from toolSpec for Claude Sonnet 4 (#31943)
* fix(bedrock): drop strict/additionalProperties from toolSpec for Claude Sonnet 4 Claude Sonnet 4 on Bedrock Converse rejects toolSpec.strict and additionalProperties the same way Opus 4.7/4.8 do. Add bedrock_converse_supports_strict_tools: false to all Sonnet 4 regional variants so those fields are suppressed before the request is sent. Co-authored-by: Cursor <cursoragent@cursor.com> * test(bedrock): assert additionalProperties dropped for strict-unsupported models Rename the regression test to reflect Opus 4.7/4.8 and Sonnet 4 coverage, and assert both strict and additionalProperties are stripped from toolSpec. Co-authored-by: Cursor <cursoragent@cursor.com> * test(fireworks): skip embeddings live test when provider account is suspended --------- Co-authored-by: Cursor <cursoragent@cursor.com> Co-authored-by: Claude <noreply@anthropic.com> |
||
|
Mateo Wang
|
b76a858826
|
feat: declarative fallback generalizations for unknown models (#29718)
* feat: declarative fallback generalizations for unknown models Unknown or newly-released models previously degraded (missed cost lookups, wrong supports_* flags, broken provider routing) and were patched with one-off hardcoded regexes scattered across Python. This adds a single data-driven source of truth: a fallback_generalizations block in model_prices_and_context_window.json holding ordered, case-insensitive regex rules that map a model name to the metadata to apply when it has no exact entry. A new fallback_generalizations module owns the rules and a compiled-regex cache that is built once and invalidated on reload, so the O(n) scan runs only on a cache miss. get_llm_provider now routes an otherwise-unknown model via the first matching rule's litellm_provider, replacing the hardcoded _CLAUDE_PATTERN and _matches_claude_model_pattern. _get_model_info_helper falls back to a matching rule's model_info after the exact lookups miss, so get_model_info and the supports_* helpers resolve unknown models from the same rule. get_model_cost_map extracts the block out of the returned map, and the integrity check now counts real model entries (excluding reserved meta keys) so the new key cannot mask a genuinely shrunk upstream file. The top level of the file stays a flat map of models so existing litellm releases that fetch the live file keep working and keep receiving updates; the block ships in both the root file and the bundled backup. An anthropic-claude rule reproduces the old future-claude routing and additionally supplies capability flags and a context window https://claude.ai/code/session_01G8Jro8dPLktwnaaSJwVDpo * refactor(anthropic): derive adaptive-thinking from a version threshold; harden generalizations Replace the per-minor-version _is_claude_4_6_model / _is_claude_4_7_model substring matchers with a single _claude_version_at_least predicate that parses the Claude family version from the model name and compares against 4.6. This covers 4.8/4.9/5.x without a code change (the old matchers missed 4.8 entirely) while keeping an explicit supports_adaptive_thinking flag authoritative when present, so there is one source of truth. The two direct call sites in the chat transformation now route through _is_adaptive_thinking_model instead of the deleted matchers. Also address review feedback on the generalizations module: return a copy of the matched model_info so a future caller cannot mutate the compiled-rule cache, document that patterns are matched with re.search and must anchor with ^ and $, and reindent the fallback_generalizations block to the file's 2-space style in both JSON files. https://claude.ai/code/session_01G8Jro8dPLktwnaaSJwVDpo * fix(anthropic): surface adaptive-thinking from the cost map; fix date misparse supports_adaptive_thinking shipped in the model cost map but was never declared on ModelInfo nor copied during construction, so get_model_info (and the supports_* factory) silently dropped it for every provider-prefixed or generalized name; only a bare base entry resolved. Wire it through ModelInfo like the other capability flags and backfill the flag onto the genuine Claude 4.6/4.7/4.8 entries across providers so the data, not code, declares the capability. The anthropic-claude fallback rule also carries the flag (and now accepts a dotted minor, e.g. 4.6) so an unmapped future Claude degrades to adaptive thinking without a code change. Tighten the Claude version parser so an eight-digit date suffix (claude-opus-4-20250514, the non-adaptive Opus 4.0) is no longer read as minor 4.20250514. The cost map stays authoritative; the version check is only a fallback for provider-prefixed names (bedrock/invoke routes, -v1-less ids) that resolve to no mapped entry and so cannot be reached by an exact lookup or the bare-name rule. https://claude.ai/code/session_01G8Jro8dPLktwnaaSJwVDpo * fix(anthropic): date-safe adaptive-thinking version fallback, conservative fallback pricing, ruff strict gate Reconcile adaptive-thinking detection after merging litellm_internal_staging. Keep the cost-map resolver (_supports_model_capability) as the source of truth and add a date-safe opus/sonnet/haiku >= 4.6 name version as a fallback for provider-prefixed ids the cost map cannot resolve (e.g. bedrock/invoke/us.anthropic.claude-opus-4-6). A two-digit cap on the minor keeps an eight-digit date suffix from being misread as a minor version, so the dated Claude 4.0 release stays non-adaptive Price the shipped anthropic-claude fallback rule at the Opus tier so an unknown or newly released Claude is over-costed rather than billed as free Drop the module-level global state in fallback_generalizations (PLW0603) in favor of a small registry object, and switch its annotations plus the new utils helper to builtin generics (UP006), bringing the ruff strict-rule totals back under ceiling * refactor(anthropic): drive adaptive-thinking version gate from a declarative rule Replace the bespoke _claude_version_at_least heuristic with a version-gated fallback_generalizations rule. Unmapped Claude ids now resolve adaptive thinking purely from the cost map: an explicit entry, or the new self-contained anthropic-claude-adaptive-thinking rule that matches opus/sonnet/haiku >= 4.6 (covering 5.x, 6.x and beyond with no code change). New families ship via Price Data Reload instead of a code edit The rule carries the same Opus-tier pricing as the broad anthropic-claude rule plus supports_adaptive_thinking, and is matched first; the broad rule stays version-neutral, so an unmapped >= 4.6 Claude resolves to full pricing and the adaptive flag from one rule, while a sub-4.6 alias such as claude-opus-4-0 is still priced yet stays non-adaptive. The regex caps the minor at two digits so a dated 4.0 id (...-4-20250514) is never read as a >= 4.6 minor * refactor(anthropic): dedupe adaptive-thinking rule via declarative extends The version-gated anthropic-claude-adaptive-thinking rule duplicated the broad anthropic-claude rule's entire Opus-tier price block because rules do not merge: first match wins and returns one rule's whole model_info, so the adaptive rule had to be self-contained. Add a declarative extends field to fallback_generalizations: a rule names a parent and inherits its model_info, with its own keys overriding. Inheritance is resolved once at install time against each rule's raw model_info, so the adaptive rule now carries only its delta (supports_adaptive_thinking) and inherits pricing from the broad rule. Runtime matching, provider routing and gating are unchanged; the broad rule stays anchored and first-match-wins still holds. * docs(anthropic): add ignored description key documenting each generalization regex * fix(anthropic): drop fabricated pricing from the anthropic-claude fallback rule Per review feedback, the base rule no longer carries input/output/cache costs, and the adaptive-thinking rule that extends it inherits that no-pricing model_info. Pricing an unmapped model at a guessed tier reports a confidently-wrong cost without the caller knowing; dropping it keeps the standard unpriced behavior (zero, not a fabricated number) so a missing price stays visible. The rules still supply provider routing, context window, and capability flags, so a brand-new Claude can still be called and its capabilities (including adaptive thinking for >= 4.6) resolved. Description and tests updated to match |
||
|
yuneng-jiang
|
63cf835b14
|
Merge pull request #31420 from BerriAI/litellm_/lucid-wilson-408605
test(pass-through): fix langfuse auth=true test broken by allowed_passthrough_routes gate |
||
|
Sameer Kankute
|
133da06aa3
|
chore: litellm oss staging (#31185)
* fix(ui): widen Y-axis gutter on Usage charts so large token/request labels aren't clipped
The Total Tokens Over Time and Total Requests Over Time AreaCharts on the
Usage page used Tremor's default yAxisWidth (~56 px), which is too narrow
once totals pass the hundred-million mark — leading digits of labels like
"100.00M" / "4500.00M" got clipped against the chart edge. The requests
chart was worse: it formatted with toLocaleString(), so billion-scale
request counts produced "1,000,000,000" (13 chars) and overflowed
immediately.
Fix in two places so neither alone has to carry the whole margin:
- activity_metrics.tsx: add yAxisWidth={80} to both AreaCharts, and
switch the requests chart to the shared valueFormatter so it uses the
same compact k/M/B suffixes as the tokens chart.
- value_formatters.tsx: add a >= 1e9 branch to valueFormatter /
valueFormatterSpend that emits a "B" suffix (4.50B, $4.50B), keeping
every formatted label at most 7 chars.
Co-Authored-By: Claude Opus 4 (1M context) <noreply@anthropic.com>
* Update ui/litellm-dashboard/src/components/UsagePage/utils/value_formatters.tsx
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
* docs(readme): add Deploy on AWS/GCP with Terraform section
Adds a quickstart for the two published Terraform modules on the public
registry (BerriAI/litellm/aws and BerriAI/litellm/google). Copy-paste
main.tf for each cloud, the one-time GCP Artifact Registry remote-repo
command, and pointers to the registry pages for the full input surface.
Sits inside the Get Started section, between the gateway/SDK table and
Run in Developer Mode -- where someone scanning the README for "how do I
deploy this" will land.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
* docs(readme): add 1-click deploy buttons for AWS + GCP
GCP gets the real 1-click: Open in Cloud Shell badge that clones the repo
and walks through `terraform apply` via the existing DeployStack
tutorial (already shipped at terraform/litellm/gcp/examples/default/
TUTORIAL.md). User just picks a project.
AWS gets a soft 1-click: a Launch in AWS CloudShell badge that opens an
in-browser, already-authenticated shell. User runs four commands
(clone + cd + cp tfvars + terraform apply) once inside. There's no
native AWS deeplink that pre-clones a repo + runs a tutorial -- CFN
"Launch Stack" + CodeBuild would be needed for that, and that's a
separate piece of work.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
* docs(readme): move AWS + GCP deploy buttons next to Render button
* docs(readme): unify deploy button sizes and badge styles
* docs(readme): bump deploy button height to 48 to match Render/Railway
* docs(readme): bump AWS/GCP badge height to compensate for SVG padding
* docs(readme): bump AWS/GCP badge height to 72
* docs(readme): bump AWS/GCP badge height to 84
* fix(readme): make deploy buttons same height (48px)
https://claude.ai/code/session_01MxQRMHSDXbqJh74rF86UBc
* docs(readme): flag GCP project ID substitution in image_registry
* docs(readme): equalize deploy button heights and fix Cloud Shell button font
GitHub rewrites an image's height attribute to "height: auto; max-height: Npx", which only caps and never stretches, so each image renders at its intrinsic height. The AWS/GCP shields badges are intrinsically 28px while the Render/Railway buttons are 40px, leaving the row uneven regardless of the height="48" we set. Replace the two shields badges with committed 40px PNGs so all four header buttons render at the same 40px.
Also swap the Cloud Shell button from open-btn.svg to open-btn.png. The SVG renders its label as live text with font-family "Roboto, Sans" and no generic fallback; since neither font exists in GitHub's render environment, the text fell back to a serif (Times New Roman). The PNG bakes in the correct typeface.
* docs(readme): collapse Railway deploy anchor to a single line
The Railway button wrapped its img across indented lines, so the anchor contained leading and trailing whitespace. GitHub underlines link content, rendering that whitespace as a small blue underline beside the button. Put the anchor on one line like the other three buttons so there is no inner whitespace to underline.
* Add Claude Fable 5 cost map entries as a data-only hotfix
Backports only the model map changes from #30064 so deployments on
released litellm versions pick up Fable 5 pricing, context window, and
the adaptive thinking flag through the hosted cost map fetch without
upgrading. Includes the supports_sampling_params flag on the 28
Fable 5 / Opus 4.7 / Opus 4.8 entries (ignored by released code, read
by the gating that ships with the next release) and the matching
one-line schema declaration so the map validation test passes.
https://claude.ai/code/session_01MZarYYT3aS7DxaNjoax6Gm
* fix: correct context window tokens for GPT-5 Pro and GPT-5.4 Mini/Nano
Three bugs in model_prices_and_context_window.json:
1. gpt-5-pro and gpt-5-pro-2025-10-06: max_input_tokens and max_tokens
were SWAPPED. GPT-5 Pro has a 400K context window (input) with 128K
max output, but the values were set as max_input=128000,
max_tokens=272000. This caused token limit errors when sending
prompts over 128K tokens to GPT-5 Pro.
2. gpt-5.4-mini and gpt-5.4-mini-2026-03-17: max_input_tokens was
272000, but GPT-5.4 Mini shares the same 1,050,000 token context
window as GPT-5.4. This was inconsistent with the azure/ variants
which already correctly had 1,050,000.
3. gpt-5.4-nano and gpt-5.4-nano-2026-03-17: same issue as Mini,
max_input_tokens was 272000 instead of 1,050,000.
Source: OpenAI model documentation and contextwindows.dev which
aggregates official context window sizes.
Fixes #30928 (partially — the issue incorrectly claims gpt-5/gpt-5-mini
should be 400K; their 272K values are correct per OpenAI docs)
* fix: also correct max_output_tokens for gpt-5-pro (272000→128000)
Per reviewer feedback, max_output_tokens was left at 272000 while
max_tokens was corrected to 128000, causing an internal inconsistency.
Both should be 128000 per OpenAI docs.
* fix(cost): price gpt-image generated output tokens as image tokens (#31147)
The OpenAI Images endpoints (/v1/images/generations, /v1/images/edits) return
usage with no output token breakdown — litellm's `ImageUsage` has no
`output_tokens_details` field — so generated-image OUTPUT tokens were priced at
the text rate (`output_cost_per_token`) instead of the image rate
(`output_cost_per_image_token`). For gpt-image-2 that is $10/1M vs $30/1M, a ~3x
undercount on the dominant cost component (image output is ~74% of spend). This
also affects azure gpt-image, which shares this calculator.
The OpenAI gpt-image cost calculator re-implemented usage handling instead of
reusing `calculate_image_response_cost_from_usage`, the shared helper that
azure_ai/gemini/vertex_ai already use. That helper classifies generated output
tokens as image tokens when the provider does not itemize output, and splits
text/image when it does.
Fix: route the ImageUsage path through `calculate_image_response_cost_from_usage`
(pre-transformed chat Usage objects are still costed directly). Adds a regression
test for the no-breakdown ImageUsage case (gpt-image-2).
* fix(bedrock): route application-inference-profile ARNs to converse (#18258) (#31098)
A bare application-inference-profile ARN passed as bedrock/arn:... fell
through to the invoke route, which cannot derive a provider from the
opaque profile id and raised 'Unknown provider=None'. The converse route
needs no provider, so detect these ARNs in get_bedrock_route and route
them to converse, matching the behavior of the already-documented
bedrock/converse/arn:... workaround.
Explicit invoke/ prefixes still win, and they remain a dead end for these
ARNs by design (no provider derivable). System-defined inference-profile
ARNs that embed a known model, and other opaque ARN types
(provisioned-model, imported-model, custom-model-deployment) that are
frequently invoke-only, are deliberately left on their current routes;
tests guard both boundaries.
* fix(moonshot): stop mutating caller messages on tool_choice='required' (#31060)
_add_tool_choice_required_message appended the "select a tool" prompt to
the caller's messages list in place, so transform_request corrupted the
caller's conversation history and appended a duplicate prompt on every
retry. Build and return a new list instead so the call stays idempotent.
Adds a regression test asserting the input messages list is unchanged
across repeated transform_request calls.
Co-authored-by: Wassbdr <wassim.badraoui07@gmail.com>
* fix(transcription): accept fractional usage.seconds in diarized_json responses (#30996)
gpt-4o-transcribe and compatible ASR backends return a diarized_json
response with usage={"type": "duration", "seconds": <float>}, e.g. 295.8.
TranscriptionUsageDurationObject typed seconds as int, so parsing the
response raised a pydantic ValidationError (int_from_float). That error
surfaces as an APIConnectionError which the router treats as retryable, so
it keeps re-calling the upstream (200 every time) until the upstream
rate-limits and returns 429 to the caller.
OpenAI specs this field as a float (see openai SDK UsageDuration.seconds),
so widen seconds to float. With the parse succeeding there is no exception
left to retry, which removes the loop.
Co-authored-by: Neimar Avila <19142978+neimaravila@users.noreply.github.com>
* fix(deepseek): drop non-function tools before chat completions call (#30910)
* fix(deepseek): drop non-function tools before chat completions call
DeepSeek's /chat/completions only accepts tools of type "function".
Requests bridged from /v1/responses can carry responses-API-native tool
types, for example a Codex CLI tool typed "namespace", which DeepSeek
rejects with "unknown variant 'namespace', expected 'function'" so the
whole request fails (issue #30722).
Filter unsupported tool types in the DeepSeek request transform so the
function tools still go through; when nothing callable remains, also drop
the now-dangling tool_choice and parallel_tool_calls
Fixes #30722
* test(deepseek): cover async tool filtering and document tool_choice assumption
Add an async_transform_request regression test so the sync and async tool
filtering paths cannot silently diverge, and document in _drop_unsupported_tools
that only non-function tools are dropped, so a function-named tool_choice always
references a surviving tool
* feat(catalog): add zai/glm-5.1, zai/glm-4.7-flash, openrouter/z-ai/glm-5.1 (#29840)
* feat(ui): surface team budget on key overview when key has no own budget (#30801)
* feat(ui): surface team budget on key overview when key has no own budget
* fix(ui): replace IIFE with derived variable and use find() for team budget display
* fix(anthropic): emit replayable streaming thinking blocks (#31022)
* feat(proxy): read cold-storage prompts back in the logs detail view (#30364)
* feat(proxy): read cold-storage prompts back in the logs detail view
When a deployment offloads prompts and responses to cold storage instead of
Postgres, the spend-log row holds only "{}" placeholders plus a
metadata.cold_storage_object_key pointer, so the UI logs detail drawer showed
nothing. The detail endpoint only read the placeholder columns and never
fetched the object back.
Resolve the payload per row based on actual content, not a config flag: if
Postgres has content, return it; otherwise read the exact stored object key and
fetch from the configured cold storage backend through ColdStorageHandler.
Reading the persisted key is a single GET. The key embeds a microsecond
timestamp that cannot be reconstructed from the millisecond-precision startTime
column, and listing the day's prefix to match on request_id would be too
expensive for this per-open path.
Also teach the detail drawer's pretty-view parser to accept a bare messages
array. The cold storage payload carries the prompt as a top-level messages list
with no proxy_server_request, so without this the output rendered while the
input stayed blank.
ColdStorageHandler gains an optional injected logger so the resolver can be unit
tested without monkeypatching. Postgres-stored prompts are unaffected: the fast
path returns the existing columns and the request-body object still renders the
same way.
* Update litellm/proxy/spend_tracking/spend_management_endpoints.py
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
* test(proxy): cover ColdStorageHandler resolution paths and cold-storage fetch failure
Add unit tests for ColdStorageHandler (injected logger, graceful None when no
logger is configured, and resolution of a configured logger from the callback
registry) and a regression test asserting a cold storage backend exception
degrades to the Postgres values instead of surfacing a 500.
---------
Co-authored-by: Bytechoreographer <Bytechoreographer@users.noreply.github.com>
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
* fix(mavvrik): advance metricsMarker after upload; fix scheduler startup (#31068)
* fix(mavvrik): advance metricsMarker after upload + fix scheduler startup
Two bugs fixed:
1. deliver() never called PATCH /metrics/agent/ai/{connectionId} after a
successful GCS upload, so metricsMarker stayed at 0 and every daily run
re-exported the same dates in an infinite catch-up loop.
Fix: add _update_metrics_marker(date_epoch) called at the end of deliver()
after _upload_to_gcs() succeeds. A 4xx warns but does not raise (the GCS
file is already committed). A 410 raises consistent with the rest of the
destination.
2. init_mavvrik_focus_background_job runs at proxy startup before any LLM call
has triggered lazy instantiation of MavvrikFocusLogger, so it found no
logger instance and silently skipped registering the daily export job.
Fix: if no instance is found but "mavvrik" is in litellm.callbacks, call
_init_custom_logger_compatible_class to force instantiation before
the APScheduler job is registered.
* fix(mavvrik): catch up from earliest window when metricsMarker=0
When the connector is freshly registered, metricsMarker=0 parses to None.
The catch-up block was guarded by `if last_ingested and ...` which skipped
it entirely for None, so only yesterday was exported instead of the full
_MAX_CATCHUP_DAYS window.
Fix: treat None as being _MAX_CATCHUP_DAYS behind (start from earliest_catchup).
The existing > 7 day warning only fires for non-None markers that are old.
* fix(mavvrik): use now as end_time for yesterday's export window
LiteLLM_DailyUserSpend rows for a given date get their updated_at
bumped by the spend flush job throughout the next morning. The core
database query filters on updated_at, so capping end_time at midnight
(yesterday + 1 day) missed any spend rows flushed after midnight.
Fix: pass now (cron fire time) as end_time for the daily "yesterday"
window so all fully-settled rows are captured regardless of when the
flush job ran.
Verified: claude-3-5-sonnet BilledCost went from 0.0 to ~$2.40 per
row in the exported FOCUS CSV.
* fix(mavvrik): also use now as end_time for catch-up windows
* fix(mavvrik_focus): pass required args to _init_custom_logger_compatible_class
Calling it with only logging_integration raised TypeError at proxy startup
because internal_usage_cache and llm_router have no defaults. Also fix test
name to reflect the actual status code (5xx not 4xx) used in the mock.
* ci: retrigger CI run
* feat: pass through optional `instruction` field in the rerank API (vLLM/Qwen3-Reranker) (#30757)
* Add optional `instruction` passthrough to the rerank API
vLLM's /v1/rerank and /v1/score accept an optional top-level `instruction`
field (folded into the model's chat_template_kwargs and consumed by the
chat template — e.g. Qwen3-Reranker). LiteLLM's managed rerank route silently
dropped it: RerankRequest / OptionalRerankParams had no such field, so the
outgoing body was rebuilt without it.
Thread an opt-in `instruction: Optional[str]` through rerank()/arerank(),
get_optional_rerank_params, and the hosted_vllm transformation into the
request body, only when non-None. When callers omit it, model_dump(exclude_none)
drops the field and the outgoing request is byte-for-byte unchanged — fully
backward-compatible. (DeepInfra already forwards `instruction` via
non_default_params; this formalizes the field in the shared types.)
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* Address review: thread `instruction` as a typed param + cover rerank_utils
Per PR review (greptile P2 + codecov):
- Make `instruction` a typed, named argument on the rerank provider interface
instead of recovering it from the opaque `non_default_params` blob. Adds
`instruction: Optional[str] = None` to `BaseRerankConfig.map_cohere_rerank_params`
and every provider override, and forwards it explicitly from
`get_optional_rerank_params`. hosted_vllm now reads the named param directly.
It is still also surfaced in `non_default_params` so providers that read it
there (e.g. DeepInfra) keep working now that `rerank()` consumes `instruction`
as a named param rather than leaving it in **kwargs.
- Add get_optional_rerank_params unit tests (present + absent) to cover the
previously-uncovered threading line flagged by codecov.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix: scan rerank `instruction` through request guardrails
The rerank guardrail translation (CohereRerankHandler.process_input_messages)
only scanned `query`, so the newly added `instruction` field reached the
backend model unscanned. Since instruction-aware rerankers (hosted vLLM /
Qwen3-Reranker) fold `instruction` into the prompt, an authenticated caller
could place content there to bypass configured rerank request guardrails.
Generalize the handler to scan every user-controlled text field (`query` and
`instruction`) in one apply_guardrail call and write each sanitized value back
by index. Query-only requests are unchanged (single-element list at index 0);
non-string fields are left untouched. Adds tests covering instruction
scanning, PII masking write-back, and the non-string case.
Addresses the Veria AI security review on PR #30757.
* test: narrow Optional results before len() to satisfy basedpyright budget
The lint gate (basedpyright delta-vs-base budget) flagged one new
reportArgumentType: len(result.results) where results is
List[RerankResponseResult] | None. Assert results is not None first to
narrow the type before len()/indexing.
* fix: read rerank `instruction` from kwargs to satisfy basedpyright budget
The basedpyright delta-vs-base gate flagged one new reportArgumentType: the
Router forwards rerank calls via an untyped `**kwargs` unpack
(`litellm.arerank(**{**data, **kwargs})`), and declaring `instruction` as a
typed named param on the public `rerank`/`arerank` entrypoints made pyright
check that key against `str | None`, adding an error at router.py with no real
safety gain. Read `instruction` from kwargs in `rerank` instead.
It remains fully typed where it matters - threaded as a typed argument through
`get_optional_rerank_params` and each provider's `map_cohere_rerank_params`
(the original Greptile P2 ask). Whole-repo reportArgumentType is back to the
base count (net 0); rerank hosted_vllm + cohere guardrail suites pass; ruff clean.
---------
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* fix(github_copilot): synthesize empty choices at the provider seam (#30929)
Newer Copilot Claude models (opus-4.7, opus-4.8) return responses with
choices=[], either carrying Anthropic-native content blocks or, for the
max_tokens=1 probe Claude Code sends, no content at all. github_copilot
is dispatched through the OpenAI SDK handler, which calls
convert_to_model_response_object directly and never invokes
GithubCopilotConfig.transform_response, so the empty-choices guard there
surfaced as a 500
Instead of synthesizing choices inside the shared
convert_to_model_response_object (which would silently turn empty choices
into a fabricated success for every provider), add a no-op
transform_parsed_response_dict hook on BaseConfig. GithubCopilotConfig
overrides it to synthesize choices from Anthropic-native content, reusing
its existing parsing, and the OpenAI SDK handler routes its parsed
response through the hook before generic conversion. The core utility
keeps treating empty choices as an error for all other providers
Fixes: https://github.com/BerriAI/litellm/issues/30927
Signed-off-by: David J. M. Karlsen <david@davidkarlsen.com>
* fix(router): stop fallback lookups from mutating the router fallbacks config (#30624)
* fix: correct amazon.titan-embed-text-v2 input price to $0.02/1M tokens (#29693)
* fix: correct amazon.titan-embed-text-v2 input price to $0.02/1M tokens
* test: scope local cost map env var with monkeypatch to avoid test pollution
* fix(sensitive_data_masker): fully mask secrets at or below the reveal threshold (#30764)
* fix(sensitive_data_masker): fully mask secrets at or below the reveal threshold
_mask_value did partial reveal by showing the first visible_prefix and last
visible_suffix characters, but for a value whose length was at or below
visible_prefix + visible_suffix (8 by default) it returned the value verbatim.
A value of exactly 8 chars fell through the length guard and computed
masked_length == 0, reconstructing the original string with no mask characters;
anything shorter hit the early return. Either way short credentials were emitted
in plaintext.
mask_dict routes real secrets through this path, so an 8-char-or-shorter redis
password, api key, or token could be written to logs and the UI unmasked. The
sibling helper mask_sensitive_keys already guards this case; _mask_value now does
the same by fully masking any value at or below the threshold.
* fix(sensitive_data_masker): add mask_short_values opt-out for truncation callers
Fully masking short values is the right default for secret masking, but
CooldownCache reuses the masker purely to truncate exception messages to the
first 50 characters, and it relies on short messages being returned readable.
Masking those blanked out short exception text and broke its tests.
Add a mask_short_values flag (default True, secure) and have CooldownCache pass
False so it keeps the truncation behavior, while every secret-masking caller
still gets short values fully masked.
* fix(mcp_debug): opt out of short-value masking to keep diagnostic token preview
MCPDebug uses the masker to preview auth tokens in debug headers and documents
that values of 10 chars or fewer are shown unchanged so token types stay
distinguishable. Pass mask_short_values=False so that diagnostic behavior is
preserved while secret maskers keep masking short values.
* fix(mcp_debug): mask short auth values in debug headers instead of echoing them
Earlier this masker opted out of short-value masking to keep a token preview, but
that echoes short authorization and token values verbatim in debug response
headers, which is the same leak this change is meant to close. Auth material
should never be emitted in full, so mask short values here too; the first/last
character preview still applies to longer tokens. Only CooldownCache keeps the
opt-out, since it truncates exception text rather than masking secrets.
* test(mcp_debug): assert masked short value preserves length
* refactor(fireworks_ai): remove deprecated audio transcriptions endpoint (#30917)
Fireworks AI deprecated audio inference on 2026-06-10
(https://docs.fireworks.ai/updates/changelog#audio-inference-and-image-generation-deprecation).
Live API testing confirms the endpoint is already non-functional: a valid
Fireworks API key receives HTTP 401 "Unauthorized" from
api.fireworks.ai/inference/v1/audio/transcriptions for every request,
regardless of payload. The audio-prod.api.fireworks.ai host referenced in
the test suite returns 401 for every path; the entire host is decommissioned.
Remove the dead FireworksAIAudioTranscriptionConfig class and every
reference to it across the codebase:
- Delete litellm/llms/fireworks_ai/audio_transcription/ directory (17-line
config class that inherited from OpenAIWhisperAudioTranscriptionConfig)
- Remove the Fireworks branch from
ProviderConfigManager.get_provider_audio_transcription_config() in
litellm/utils.py; update the stale comment in
get_optional_params_transcription that referenced fireworks ai
- Remove the FireworksAIAudioTranscriptionConfig entries from
LLM_CONFIG_NAMES and _LLM_CONFIGS_IMPORT_MAP in
litellm/_lazy_imports_registry.py
- Remove the TYPE_CHECKING re-export in litellm/__init__.py
- Remove the transcription branch in the fireworks_ai case of
get_supported_openai_params() in
litellm/litellm_core_utils/get_supported_openai_params.py
- Remove the whisper-v3 and whisper-v3-turbo entries from
model_prices_and_context_window.json and
litellm/model_prices_and_context_window_backup.json (both had
mode: audio_transcription and zero-cost pricing)
- Remove the TestFireworksAIAudioTranscription test class and its
imports from tests/llm_translation/test_fireworks_ai_translation.py
No other provider is affected. The openai_compatible_providers list,
FireworksAIMixin, and the OpenAI Whisper transcription handler all stay
because they are shared with other Fireworks endpoints and other
providers. The provider_endpoints_support.json registry already had
audio_transcriptions set to false for fireworks_ai.
* feat: add darkbloom provider (#30876)
* feat: add darkbloom provider
* fix: document darkbloom provider endpoints
* fix: address darkbloom review feedback
* fix: update darkbloom tool metadata
* fix: fail fast for non-Postgres database URLs (#30883)
* fix(proxy): fail fast on non-PostgreSQL DATABASE_URL instead of hanging on startup
LiteLLM's Prisma datasource is pinned to provider = 'postgresql', so a sqlite:// or mysql:// DATABASE_URL can never connect.
Today that surfaces as an opaque startup stall where the port never binds, and a separate 'DB not connected' 500 on /key/generate when no DATABASE_URL is set at all leaves operators guessing what to configure.
Validate the DATABASE_URL / DIRECT_URL scheme in run_server before any Prisma call and exit with an actionable message naming the unsupported scheme.
Also reword CommonProxyErrors.db_not_connected_error to tell the operator to set DATABASE_URL to a postgresql:// connection string.
Add regression tests covering postgres acceptance and sqlite/mysql/mssql rejection.
* fix: resolve CI failures and proxy DB URL typing issue
* fix(proxy): fail fast on non-PostgreSQL DATABASE_URLs with clear startup errors instead of hanging
* Validate DIRECT_URL alongside DATABASE_URL startup guards
* fix(bedrock): surface modeled HTTP status for mid-stream error events so 5xx is retryable (#24608) (#30946)
* fix(bedrock): surface modeled HTTP status for mid-stream error events (#24608)
* test(bedrock): mid-stream server errors trigger streaming fallback (#24608)
* style(bedrock): black-format stream-error helper (#24608)
* fix(mcp): re-land native tool preservation with typed annotations (#30645)
* fix(mcp): preserve native tools in semantic filter hook with typed annotations
* fix(mcp): tighten _is_mcp_tool Chat Completions shape check
* fix(sambanova): return embeddings supported params instead of dropping them (#30937)
* fix(router): send fallback metadata when streaming (#30914)
When a streaming request triggers a fallback, there was previously no way to
know it happened. This commit addresses this in a few ways:
1. The response now correctly populates the fallback headers
(`x-litellm-attempted-fallbacks`) so callers know a fallback happened.
2. The correct model ID is passed in the streaming chunks.
3. A streaming chunk with the fallback error can be optionally sent back
to the client (opt-in) by passing `include_fallback_errors: true` in
the request.
The format of the fallback errors while streaming is intentionally OpenAI
compatible to not break existing libraries that parse these events. It was
tested with Vercel's AI SDK (ai-sdk.dev). It is also opt-in, so it is not
delieved unexpectedly to callers by default.
* fix(mistral): drop output-only reasoning fields from input messages (#30884)
LiteLLM attaches reasoning_content and thinking_blocks to assistant
responses. Replaying those assistant turns verbatim forwarded the fields
back to Mistral, whose input schema forbids unknown keys, so the whole
request failed with a 422 extra_forbidden and reasoning models became
unusable across multiple turns.
Strip both fields from assistant messages before the request is built, in
a spot that runs ahead of the image/file branch so it applies on every
path. Fixes #30835
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(perplexity): bill search queries at the per-request price, not 1/1000 of it (#30652)
* fix(perplexity): bill search queries at the per-request price, not 1/1000
The fallback cost calculator divided search_context_cost_per_query by
1000, but that field stores the per-request price in USD: sonar is
{low: 0.005, medium: 0.008, high: 0.012}, matching Perplexity's published
$5/$8/$12 per 1,000 requests expressed per request. The gemini cost
calculator reads the same field per request with no division (its
docstring calls it "the per-request cost").
The division understated search cost by 1000x on every Perplexity call
that falls back to manual calculation (i.e. when the API does not return
a pre-computed usage.cost). Use the value directly.
Update the tests that had encoded the /1000 factor in their expectations,
and drop an unused import flagged by ruff in the touched test file.
* test(perplexity): update integration test search-cost expectations to per-request
The integration tests still encoded the old /1000 search-cost factor, so
they failed once the fallback calculator was corrected to bill
search_context_cost_per_query per request. Update the four expected-cost
computations (and the high-volume dollar-value comments) to match.
* test(perplexity): drop unused mock imports flagged by ruff
* fix: include model_access_groups when expanding all-team-models in get_team_models (#30622)
* fix(fireworks_ai): return None for transcription in get_supported_openai_params
Fireworks AI deprecated audio inference on 2026-06-10; the endpoint is
decommissioned. Without an explicit transcription branch, requests with
request_type='transcription' fell through to the else and returned
FireworksAIConfig chat-completion params. Return None instead to signal
the provider does not support transcription.
* fix(proxy): gate include_fallback_errors behind expose_fallback_errors_to_caller setting
Without an operator gate, any authenticated caller could set include_fallback_errors=True,
trigger a fallback, and read raw upstream exception messages from the
x-litellm-fallback-errors header and the litellm-fallback-metadata SSE event.
Strip include_fallback_errors from request data in common_processing_pre_call_logic
when expose_fallback_errors_to_caller is not set, so the router never builds the
error list. Also gate _should_include_fallback_errors on the same setting as a
secondary check for the streaming SSE injection path.
* test(proxy): opt in to expose_fallback_errors_to_caller in streaming SSE test
The operator gate added in
|
||
|
Yuneng Jiang
|
432d99f3ee
|
test(pass-through): grant allowed_passthrough_routes so langfuse auth=true test reaches rpm path
#29256 made auth=true pass-through routes deny-by-default unless the key/team has allowed_passthrough_routes configured, but this integration test was not updated. The test key had no allowlist, so the auth=true parametrizations (rpm_limit=0 -> expect 429, rpm_limit=2 -> expect 207) now hit the 403 gate in auth before reaching the rpm/forwarding logic they mean to exercise. Grant the test key allowed_passthrough_routes for /api/public/ingestion so it clears the gate. Also removes a latent order-dependency: the case only passed locally when an earlier (auth=false) parametrization registered the route first; under worker isolation (CI xdist) it failed with 403. |
||
|
Mateo Wang
|
b16cfd7de9
|
test: point router/completion/triton tests at the local fake OpenAI endpoint (#30900)
* test: point router/completion/triton tests at the local fake OpenAI endpoint The shared Railway-hosted mock (exampleopenaiendpoint-production.up.railway.app) takes down unrelated CI jobs whenever it is unreachable. #30695 moved the mounted proxy configs onto a job-local fake server but left these in-Python api_base literals pointing at the dead host, so litellm_router_testing, local_testing_part1, local_testing_part2 and llm_translation_testing still fail with a 404 "Application not found" when Railway is down Resolve the api_base from FAKE_OPENAI_API_BASE (default http://127.0.0.1:8190) through a shared helper, auto-start the canned server from the local_testing and llm_translation conftests when nothing is already serving, and extend the server with a Triton embeddings route and a slow-endpoint delay so the triton and latency-timeout tests run fully offline. The deliberately broken fallback URL is left as-is so fallback handling still has a failing upstream * fix: ignore non-loopback FAKE_OPENAI_API_BASE so the local mock is used in CI * fix: drop 0.0.0.0 from loopback hosts, an unreliable client connect target * fix(tests): keep fake OpenAI mock alive across xdist workers ensure_fake_openai_endpoint registered atexit on the worker that spawned the subprocess, so under -n 4 the first worker to drain its queue would terminate the shared mock while siblings were still hitting it. Detach the child via start_new_session and drop the per-worker teardown; reuse on /health handles re-runs and CI containers clean up themselves |
||
|
Mateo Wang
|
556e8f89c8
|
ci: run a local fake OpenAI endpoint instead of the shared Railway mock (#30695)
Several CI jobs run the proxy against a model whose api_base is a shared "fake OpenAI endpoint" hosted on Railway (exampleopenaiendpoint-production.up.railway.app) so the E2E runs return canned responses without paying for or depending on a live provider. When that single deployment is down, every one of those jobs fails with "404 Application not found" even though nothing in the PR is broken; the whole repo is coupled to the uptime of one free external service. This adds tests/_fake_openai_endpoint_server.py, a small canned-response OpenAI-shaped server (chat, text, embeddings, streaming with usage, and the "429" rate-limit special case), and a reusable start_fake_openai_endpoint CircleCI command that runs it on host port 8190 and waits until healthy. The affected jobs now inject FAKE_OPENAI_API_BASE pointing at the local server, and the example configs they mount resolve api_base from that env var. The intentionally bad fallback URL in proxy_server_config.yaml is left untouched so the fallback test still exercises a failing upstream. Wired into build_and_test, litellm_router_testing, db_migration_disable_update_check, proxy_logging_guardrails_model_info_tests, proxy_spend_accuracy_tests, proxy_multi_instance_tests, proxy_store_model_in_db_tests, and proxy_build_from_pip_tests. |
||
|
ryan-crabbe-berri
|
b5fcd859be
|
fix(guardrails): return 400 not 500 when AIM blocks a request (#30573)
* fix(guardrails): return 400 not 500 when AIM blocks a request AIM guardrail blocks raised a bare HTTPException whose type and param serialized as the literal string "None", which broke OpenAI-SDK error parsing for downstream consumers. Switching AIM to raise a ProxyException surfaced a second bug: the shared error funnel re-derived the HTTP status from a nonexistent status_code attribute and downgraded the 400 to a 500. The funnel now honors an already-normalized ProxyException rather than rebuilding it, and ProxyException is excluded from llm_exceptions alerting so a content-policy block no longer pages on-call as an LLM API failure Resolves LIT-3751 * fix(guardrails): route all AIM rejection paths through ProxyException The block-action fix left two AIM rejection paths raising a bare HTTPException: the multimodal anonymize rejection and the output-side block. Both serialized type and param as the literal string "None", the same malformed shape the block fix removed. Funnel all three through a shared _rejection helper so they return a conformant OpenAI error body. The output block carries content_policy_violation; the multimodal rejection stays a plain invalid_request_error because it is a usage error, not a policy violation Resolves LIT-3751 * fix(guardrails): record AIM ProxyException blocks in failure logs Switching AIM blocks from HTTPException to ProxyException made _is_proxy_only_llm_api_error return False for them, so _handle_logging_proxy_only_error was skipped and the blocked prompt was dropped from the configured failure loggers. Classify ProxyException as a proxy-only error alongside HTTPException so guardrail blocks are recorded again, matching the prior behavior. The llm_exceptions alert suppression is a separate check and stays in place Resolves LIT-3751 * style(guardrails): use str | None over Optional[str] in AIM _rejection * style(guardrails): collapse AIM _rejection signature per black |
||
|
Sameer Kankute
|
079c136742
|
chore(oss): litellm oss staging 120626 (#30292)
* feat(bedrock): add bedrock mantle gemma 4 models (#30264) * feat(bedrock): add bedrock mantle gemma 4 models * test(bedrock): harden mantle local cost fixture * feat(responses): enable the responses API for the Tensormesh provider (#30209) * feat(responses): enable the responses API for the Tensormesh provider * Update litellm/llms/openai_like/providers.json Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com> --------- Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com> * fix(langfuse_otel): mark LLM spans as generations (#30250) * fix(bedrock): stop stream_chunk_size leaking into invoke request bodies (#30240) stream_chunk_size is a LiteLLM-internal knob for re-chunking the HTTP response stream. The invoke transformations splat optional_params into the provider request body without dropping it, and Bedrock rejects unknown fields, so any bedrock/invoke request that sets the parameter fails with ValidationException: stream_chunk_size: Extra inputs are not permitted. Drop it in the invoke dispatcher (covers cohere, titan, mistral, meta, ai21) and in the Claude messages-format request builder (the route used for bedrock/invoke Anthropic models) * fix(bedrock): stop buffering streamed tool-call argument deltas (#30231) * fix(bedrock): stop buffering streamed tool-call argument deltas Two issues made Bedrock tool-use streaming arrive as a single end-of-stream burst through LiteLLM while plain text streamed fine. First, the anthropic-beta allowlist mapped fine-grained-tool-streaming-2025-05-14 to null for bedrock and bedrock_converse, so the header was silently stripped. Without that beta, Anthropic models on Bedrock buffer tool input server-side and emit all toolUse.input deltas at once (verified against converse-stream and invoke-with-response-stream directly). Bedrock accepts the beta via additionalModelRequestFields.anthropic_beta, so it is now forwarded. Second, the streaming reads re-chunked the AWS event stream with iter_bytes(chunk_size=1024). httpx's ByteChunker only releases full 1024-byte blocks, so the small early events (messageStart, contentBlockStart, first deltas) sat in the buffer until enough bytes accumulated, pushing time-to-first-byte from ~1.4s to ~8.5s on buffered tool-use streams. The default is now no re-chunking; an explicit stream_chunk_size is still honored. * test(bedrock): cover explicit stream_chunk_size on sync invoke path * test(bedrock): cover stream_chunk_size plumbing through converse completion * test(bedrock): cover stream_chunk_size default in legacy BedrockLLM streaming * test(bedrock): merge converse handler tests into existing mapped test file pytest imports test modules by basename in non-package test dirs, so the new tests/test_litellm/llms/bedrock/chat/test_converse_handler.py collided with the pre-existing tests/test_litellm/llms/chat/test_converse_handler.py and broke collection in CI. Move the new tests into the existing file * feat(otel): emit v2 cost breakdown + stamp tracer scope version (#30156) Read the StandardLoggingPayload cost_breakdown into a typed LLMCost on LLMCallSpanData and emit each component under litellm.cost.* (absent components omitted, so spans stay sparse). Stamp litellm.__version__ as the instrumentation scope version so every v2 span carries a deterministic scope.version. Tests under tests/test_litellm/integrations/otel/. * fix(proxy): cancel in-flight upstream LLM request on client disconnect (opt-in) (#30223) * fix(proxy): cancel in-flight upstream LLM request on client disconnect (opt-in) On the non-streaming path, base_process_llm_request awaited the LLM call with no disconnect monitoring; when the HTTP client went away the upstream request kept running until completion or request_timeout (6000s default), holding a backend slot (e.g. a vLLM GPU slot) for output nobody would read Add an opt-in general_settings.cancel_on_disconnect flag, default off, so the default code path is unchanged. When enabled, a receive-based watcher task observes http.disconnect and cancels the asyncio.gather driving the upstream call. The resulting CancelledError is converted to HTTPException 499 only when the disconnect event is set, so server-initiated cancellations still propagate as-is. The 499 then flows through _handle_llm_api_exception like any other failure, meaning post_call_failure_hook still releases max_parallel_requests slots and fires spend and alerting callbacks; it is logged at info level instead of a full traceback Also removes the dead check_request_disconnection helper in proxy_server.py (zero call sites) along with its behavior-pin tests Builds on the receive-based design from #25776 Addresses #13774. Re-fixes #22805 (regressed after the #14295 revert) Co-authored-by: CreateRandom <18438707+CreateRandom@users.noreply.github.com> * fix(proxy): scope 499 quiet logging to disconnects and harden watcher Address the two P2 findings from the Greptile review on #30223. The info-level logging in _log_llm_api_exception now applies only to the disconnect-specific HTTPException (status 499 plus the shared _CLIENT_DISCONNECT_DETAIL message), so any other 499 raised by hooks or guardrails keeps its full traceback. The disconnect watcher now catches exceptions from request.receive() (e.g. a transport reset) and logs a warning instead of dying silently, making the degradation to no-op visible; a test pins that the LLM call is not cancelled in that case --------- Co-authored-by: kursad <kursad.lacin@brado.net> Co-authored-by: CreateRandom <18438707+CreateRandom@users.noreply.github.com> * fix(bedrock): grant aws-external-anthropic:* in OIDC session policy for claude_platform (#30200) (#30205) The inline STS session policy passed to assume_role_with_web_identity acts as an IAM PERMISSION CEILING — effective permissions are the intersection of the role's identity policies and this policy. Any action not listed is silently denied even when the IAM role grants it. #27678 added the bedrock/claude_platform/<model> route but its service-side action namespace is aws-external-anthropic:*, not bedrock:*. Without a matching statement here, every claude_platform request via OIDC (GCP federation, EKS Pod Identity webhook, etc.) 403s with 'no session policy allows the aws-external-anthropic:CreateInference action' — even with a fully permissive identity policy. Add a second ClaudePlatformLiteLLM statement covering CreateInference, CreateBatchInference, CancelBatchInference, DeleteBatchInference, CountTokens, Get*, List*. Keep aws:SecureTransport=true parity with the bedrock statement. Static creds + IRSA flow through different code paths and are not affected. Fixes #30200 * fix(proxy): set Retry-After header on RouterRateLimitError 429 responses (#30098) * Set Retry-After header on RouterRateLimitError responses When all deployments for a model are in cooldown, the proxy returns a 429 whose cooldown timing is only available by parsing the error message string. RouterRateLimitError already carries cooldown_time, so expose it as a standard retry-after header in _handle_llm_api_exception. The value is rounded up so clients never retry before the cooldown window ends. Fixes #27823. * Set Retry-After after response-headers hook so cooldown wins The cooldown-derived retry-after was assigned before the post_call_response_headers_hook merge, so a callback returning a retry-after key (including a stale or empty value) silently clobbered it. Move the RouterRateLimitError block after the callback merge so the cooldown value is authoritative for this error type. * fix(router): route aspeech through async_function_with_fallbacks (#30104) * fix(router): route aspeech through async_function_with_fallbacks Router.aspeech selected a deployment and awaited litellm.aspeech directly, so TTS requests got no retry on failure and no failover to backup deployments; the except block only fired an exception alert and re-raised. Every other router endpoint (acompletion, aembedding, atranscription, arerank) already delegates to async_function_with_fallbacks Mirror the atranscription pattern: move deployment selection and the litellm.aspeech call into a private _aspeech method, then have the public aspeech set kwargs["original_function"] = self._aspeech and await self.async_function_with_fallbacks(**kwargs). _aspeech also picks up the shared _get_async_openai_model_client helper and the same total/success/fail call accounting the sibling endpoints use Fixes #27778. * fix(router): apply deployment kwargs and rpm semaphore in _aspeech Bring _aspeech fully in line with _atranscription: call _update_kwargs_with_deployment so deployment metadata, model_info, timeout, and default litellm params flow into the request, and wrap the litellm.aspeech call with the max_parallel_requests semaphore plus async_routing_strategy_pre_call_checks so TTS respects rpm limits the same way the other router endpoints do Also add a unit test that exercises _aspeech directly and asserts the deployment metadata reaches the underlying call * fix(slack_alerting): stop false-positive hanging request alerts for requests below the alerting threshold (#30106) * fix(slack_alerting): skip hanging request alerts below the threshold The hanging request check alerted on any cached request whose completion status was not yet recorded, with no minimum age check. Since the background loop runs every alerting_threshold / 2 seconds, any request that happened to be in flight at a check fired a "hanging - Ns+ request time" alert even if it was only seconds old, producing a steady stream of false positives. Add a created_at timestamp to HangingRequestData, stamped when the request enters the hanging request cache, and skip requests younger than alerting_threshold without evicting them, so a later check can still alert if they never complete. Extend the cache TTL from threshold + 60s to 1.5x threshold + 60s; with the age check, entries only become alertable after threshold seconds, and the check period is threshold / 2, so the old TTL could evict a genuinely hanging request before any check saw it cross the threshold. Fixes #27855. * fix(slack_alerting): alert once per hanging request The min-age gate stops false positives for young in-flight requests, but a genuinely hanging request still re-alerted on every checker tick within the cache TTL. With the wider TTL (1.5x threshold + 60s) that is 1-2 extra Slack notifications per stuck request at the default 600s threshold. Flag a HangingRequestData entry as alerted once its alert fires and skip flagged entries on later ticks, so each hang produces exactly one alert. The cache reference is mutated in place, so the TTL is untouched and still handles cleanup. Adds a regression test asserting one alert across multiple ticks. Fixes #27855. * fix(health): treat all-proxy-models keys as unrestricted in /health (#30087) * fix(health): treat all-proxy-models keys as unrestricted in /health A key granted all model permissions stores the literal "all-proxy-models" marker in its models list. The /health access filter compared that marker against real model_names, so the model list filtered down to nothing and the WebUI health check returned healthy_count=0, unhealthy_count=0 with HTTP 503. Skip the filter (both the live path and the background-cache model_id scoping) when the marker is present, matching how auth_checks treats SpecialModelNames.all_proxy_models. Fixes #29744. * fix(health): resolve all-team-models sentinel to the team allowlist Same failure shape as the all-proxy-models case: a key carrying the literal "all-team-models" entry matches no real model_name, so the /health access filter would zero out the model list. Resolve the sentinel to the key's team models when team_id is set, matching get_key_models in model_checks.py. Without a team_id the sentinel stays unresolved and matches nothing, denying rather than widening access, mirroring _resolve_key_models_for_auth_check. * feat(proxy): auto-enable drop_params for Claude Code requests (#30218) * feat(proxy): auto-enable drop_params for Claude Code requests Claude Code identifies itself with a claude-cli/<version> user agent and sends Anthropic-specific params (top_k, thinking, etc.) on every request. When the proxy routes those requests to a non-Anthropic provider, the unsupported params fail the call unless drop_params is configured. Detect the Claude Code user agent in add_litellm_data_to_request and default drop_params to true for those requests, without overriding an explicit drop_params value sent by the caller. * feat(proxy): respect operator litellm_settings drop_params over Claude Code default An explicit drop_params in the operator's litellm_settings (true or false) now suppresses the Claude Code user agent default, so an operator who deliberately configured drop_params: false keeps strict param validation for Claude Code clients too. The auto-default only fills the gap when neither the request body nor the config sets a value. * fix(snowflake): migrate to native endpoints with auto-routing for Claude models (#29964) * fix(snowflake): migrate to native Cortex REST API endpoints Replaces the legacy /api/v2/cortex/inference:complete endpoint with the native OpenAI-compatible /api/v2/cortex/v1/chat/completions endpoint, fixing error 390142 (Incoming request does not contain a valid payload) when using model: snowflake/<model> in LiteLLM proxy. Changes: - litellm/llms/snowflake/chat/transformation.py: route to native /cortex/v1/chat/completions, remove Snowflake-specific tool_spec payload transformation, remove content_list response handling, add stream to supported params - litellm/llms/snowflake/anthropic/transformation.py (new): SnowflakeCortexAnthropicConfig routes Claude models to /cortex/v1/messages with anthropic-version header and Anthropic->OpenAI response transform - tests: 29 unit tests covering URL routing, auth headers, payload format, and response parsing * fix(snowflake): map max_tokens to max_completion_tokens for native endpoint * fix: handle multi-turn tool conversations and OpenAI→Anthropic tool format conversion - _extract_system_and_messages now preserves tool_calls from assistant messages and converts them to Anthropic tool_use content blocks - tool role messages are converted to user role with tool_result content blocks (as required by Anthropic Messages API) - Added _transform_tools_to_anthropic() to convert OpenAI tool format (type/function/parameters) to Anthropic format (name/input_schema) - Added comprehensive tests for multi-turn tool conversations Addresses review feedback on PR #29964 * test: add coverage for malformed JSON and non-string tool arguments * fix(tests): update chat transformation tests for native OpenAI-compatible endpoint * style: apply black formatting * fix: resolve mypy type errors in anthropic transformation * fix: correct mypy type: ignore error codes (attr-defined) * fix: use max_tokens instead of max_completion_tokens for Snowflake endpoint compatibility * refactor: merge Anthropic config into unified SnowflakeConfig with auto-routing - Remove separate SnowflakeCortexAnthropicConfig and anthropic/ directory - SnowflakeConfig now auto-routes based on model name: - Claude models → /messages endpoint (Anthropic format) - All others → /chat/completions endpoint (OpenAI format) - No new provider needed (stays as SNOWFLAKE = 'snowflake') - Tool message transformation for Claude: tool_calls → tool_use blocks, tool role → user with tool_result - OpenAI → Anthropic tool format conversion (parameters → input_schema) - Addresses Greptile feedback about unwired SnowflakeCortexAnthropicConfig * fix: use max_completion_tokens for /chat/completions (Snowflake deprecated max_tokens on this endpoint) * fix(tests): update assertions for Claude auto-routing to /messages endpoint * fix(snowflake): add tool_choice conversion and preserve max_completion_tokens in Anthropic path * fix(snowflake): use ChatCompletionMessageToolCall objects and strip model prefix on OpenAI path * fix(snowflake): collect multiple system messages to prevent guardrail override * chore: remove committed .pyc files and add __pycache__ to .gitignore * fix: remove unused Union import * fix: restore original .gitignore (accidentally replaced in earlier commit) * feat(snowflake): add streaming response handler for both Anthropic and OpenAI SSE formats * fix: remove unused AsyncIterator and Iterator imports * fix: add missing total_tokens to ChatCompletionUsageBlock * fix(snowflake): coalesce consecutive tool results into single user message for Anthropic * fix(snowflake): handle message_start event for streaming input_tokens tracking * fix: evict last deleted model in multi-instance deployments (#28608) * fix: evict last deleted model in multi-instance deployments _delete_deployment had an early return when db_models was empty, preventing eviction of the last deleted model during reconciliation. - Remove len(db_models)==0 early return from _delete_deployment - Return None (not []) from _get_models_from_db on DB failure so callers can distinguish a transient failure from a genuinely empty DB - Guard _update_llm_router against None to skip updates on DB failure Fixes #28443 * test: remove dead MagicMock assignment in type_mismatch test * fix: update test to pass [] not None to _update_llm_router test_ProxyConfig__update_llm_router_bad_proxy_logging_raises was passing None as new_models to get through to the proxy_logging_obj check, but the None guard we added now returns early before reaching that path. Pass [] instead so the test exercises the intended AttributeError case. Signed-off-by: Rudra Dudhat <contact.rdudhat@gmail.com> * chore: regenerate API types to sync schema.d.ts with proxy OpenAPI spec Signed-off-by: Rudra Dudhat <contact.rdudhat@gmail.com> --------- Signed-off-by: Rudra Dudhat <contact.rdudhat@gmail.com> * fix: invalidate Redis spend counter on /key/reset_spend (#29694) * fix: set Redis spend counter to reset_to value on /key/reset_spend Previously, the Redis spend counter was always set to 0.0 after a reset, even when reset_to was a non-zero value (partial reset). This caused the budget to be under-enforced for up to 60 seconds until the counter expired and fell through to the DB. Now the counter is set to the actual reset_to value, so partial resets are reflected correctly and budget enforcement is consistent. * test: update reset_key_spend test to match direct cache set The implementation now sets spend_counter_cache directly instead of calling _invalidate_spend_counter. Update the test to verify the in_memory_cache.set_cache call with the correct key, value, and ttl. --------- Co-authored-by: michaelxer <michaelxer@users.noreply.github.com> * fix: add scaleway models pricing (#27659) * fix: Add embeddings support for Scaleway provider * fix: resolve merge conflicts * fix(main): clarify backend route handling for Swagger static assets (#30196) * fix(main): clarify backend route handling for Swagger static assets * fix(allowlist): add BACKEND_MOUNT_PATHS for Swagger static assets * fix(voyage): route multimodal embeddings to correct endpoint (#30193) * fix(voyage): route multimodal embeddings to correct endpoint * test(voyage): cover multimodal embedding edge cases * test(voyage): cover api key fallback * fix(voyage): raise early on missing api key and malformed image url * test(voyage): cover utils routing and helper * fix(voyage): route supported openai params for multimodal models * style: apply black formatting * fix(ui): infer Azure API version from API base (#30204) * fix(ui): infer Azure API version from API base * fix(ui): address Azure API version feedback * Update litellm/llms/snowflake/chat/transformation.py Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com> * feat(datadog): add team-scoped Datadog callback support (#29947) Enable teams to configure their own Datadog credentials via POST /team/{team_id}/callback, following the same pattern as Langfuse. * Merge pull request #29528 from aanchal22/litellm_byok-alias-merge fix(proxy): atomic merge for team model aliases and team.models on BYOK create * feat: add EmpirioLabs as an OpenAI-compatible provider (#30278) Co-authored-by: Adam Dalloul <adam.d.developer@gmail.com> * fix: resolve failing tests and lint in snowflake/team endpoints - Black-format snowflake/chat/transformation.py to fix lint failure - Update Anthropic config test to expect default max_tokens of 4096 (matches implementation) - Add AsyncMock + execute_raw mock to team_model_add cache-refresh pin test - Add model_dump mock and patch cache/logging in test_uses_atomic_array_append_with_dedup Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(test): update test_db_error_new_model_check for new _delete_deployment logic _delete_deployment no longer short-circuits on empty db_models — it now treats [] as a valid empty-DB state and proceeds to check config models. Mock get_config to return the two router deployments so they appear in combined_id_list and are protected, which matches the real-world scenario where a DB error occurs but the models are config-backed. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * feat(proxy): register cancel_on_disconnect in ConfigGeneralSettings and config list (#30295) * feat(proxy): register cancel_on_disconnect in ConfigGeneralSettings and config list Follow-up to #30223 per maintainer review: documents the flag in ConfigGeneralSettings with a short description and adds it to allowed_args in get_config_list so the UI and /config/list expose it. A test pins that /config/list returns the field with type Boolean, which requires both registrations to be present * chore(ui): regenerate schema.d.ts for cancel_on_disconnect --------- Co-authored-by: kursad <kursad.lacin@brado.net> * fix(datadog): never fall back to env DD_API_KEY for caller-supplied destinations Team/key-scoped Datadog loggers could be pointed at an arbitrary dd_agent_host or dd_site while omitting dd_api_key, causing the proxy's global DD_API_KEY to be sent as the DD-API-KEY header to that destination. Gate the env-var fallback behind an allow_env_credentials flag, set to False when the destination is caller-supplied, mirroring the existing langfuse/langsmith pattern. --------- Signed-off-by: Rudra Dudhat <contact.rdudhat@gmail.com> Co-authored-by: Emerson Gomes <emerson.gomes@thalesgroup.com> Co-authored-by: daitran-tensormesh <dai@tensormesh.ai> Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com> Co-authored-by: Muspi Merol <me@promplate.dev> Co-authored-by: fangkang <fangkangm@gmail.com> Co-authored-by: Chris Hoogeboom <chris.hoogeboom@gmail.com> Co-authored-by: kursadlacin <kursadlacin@gmail.com> Co-authored-by: kursad <kursad.lacin@brado.net> Co-authored-by: CreateRandom <18438707+CreateRandom@users.noreply.github.com> Co-authored-by: hcl <chenglunhu@gmail.com> Co-authored-by: Filippo Menghi <113345637+Cyberfilo@users.noreply.github.com> Co-authored-by: Mateo Wang <277851410+mateo-berri@users.noreply.github.com> Co-authored-by: sfc-gh-nashukla <navnit.shukla@snowflake.com> Co-authored-by: Rudra Dudhat <contact.rdudhat@gmail.com> Co-authored-by: Michael <52305679+michaelxer@users.noreply.github.com> Co-authored-by: michaelxer <michaelxer@users.noreply.github.com> Co-authored-by: Quentin Champenois <26109239+Quentinchampenois@users.noreply.github.com> Co-authored-by: mauriceberentsen <mauriceberentsen@live.nl> Co-authored-by: lost9999 <56498264+lost9999@users.noreply.github.com> Co-authored-by: GaetanVDB07 <86427581+GaetanVDB07@users.noreply.github.com> Co-authored-by: Aanchal Khandelwal <aan2210khandelwal@gmail.com> Co-authored-by: Adam Dalloul <adam_dalloul@icloud.com> Co-authored-by: Adam Dalloul <adam.d.developer@gmail.com> Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> |
||
|
Sameer Kankute
|
02bce7b393
|
fix(mcp): honor server_id for REST tool calls with shared upstream URLs (#30184)
* fix(mcp): honor server_id for REST tool calls with shared upstream URLs When multiple MCP server entries point at the same backend URL and tool name, REST /mcp-rest/tools/call now routes and applies auth from the requested server_id instead of the global unprefixed tool-name mapping. Co-authored-by: Cursor <cursoragent@cursor.com> * fix(mcp): classify prefixed REST tool names against full registry Use all registered MCP server prefixes for prefix detection so unauthorized prefixed names still trigger tool_server_mismatch, and reject ambiguous hyphenated REST tool names with server_id. Co-authored-by: Cursor <cursoragent@cursor.com> * test(mcp): cover server_id fallback for unresolved prefixed REST tool names execute_mcp_tool left the prefix-retry and requested-server fallback branches uncovered, dropping diff coverage below the project target. Add a regression test for a REST call that passes server_id with a prefixed tool name that resolves to no managed tool; it must still dispatch to the server identified by server_id rather than the server named by the prefix. * test(mcp): scope global tool-name mapping mutation with patch.dict * test(mcp): cover server_id guard on prefix-retry tool resolution The prefix-retry branch in execute_mcp_tool re-prefixes the tool name with the requested server's known prefixes when the bare lookup misses. The candidate-found path that assigns mcp_server from that lookup stayed uncovered, so codecov patch coverage remained below the diff target. Add a regression test where the re-prefixed lookup resolves a server whose server_id differs from the requested server_id; the tool_server_mismatch 403 guard must still fire instead of being silently bypassed. * test(mcp): assert requested server credentials injected on cross-server REST routing * perf(mcp): scan registry prefixes only when server_id is supplied * fix(mcp): allow hyphenated upstream tool names when REST server_id is authoritative * perf(mcp): skip registry prefix scan for separator-free REST tool names * test(http_handler): drop httpbin dependence from per-request timeout test The per-request timeout test posted to https://httpbin.org/delay/10 and asserted a Timeout was raised. httpbin's free /delay endpoint intermittently returns 503 even when the /get reachability guard succeeds, so local_testing_part1 flaked on that 503 instead of the expected timeout (failed identically across an initial run and a rerun-from-failed). Serve the slow response from a local ThreadingHTTPServer so the timeout fires deterministically with no third-party network dependence. --------- Co-authored-by: Cursor <cursoragent@cursor.com> Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> |
||
|
Sameer Kankute
|
7519e37d26
|
feat(proxy): enforce key/team guardrails on bedrock passthrough routes (#30194)
* feat(proxy): enforce key/team guardrails on bedrock passthrough routes
/bedrock/... passthrough routes silently skipped all guardrail hooks because
CallTypes.allm_passthrough_route had no entry in guardrail_translation_mappings.
Add a dispatcher (LlmPassthroughRouteHandler) registered for that call type that
routes to BedrockPassthroughGuardrailHandler for Bedrock Converse endpoints; wire
post_call_success_hook into both the JSON and AWS event-stream response paths in
common_request_processing, including full de-anonymization for streaming responses
with proportional text distribution across original event-stream delta frames.
* refactor(proxy): address greptile feedback on bedrock passthrough guardrails
Move botocore event-stream logic from proxy/ to BedrockPassthroughGuardrailHandler.de_anonymize_converse_stream; _handle_event_stream_allm_passthrough_route becomes a thin provider dispatcher. Pass custom_headers through _handle_non_streaming_allm_passthrough_route so early-return guardrail responses include x-litellm-call-id and related headers.
* style: run black on handler and common_request_processing
* refactor(proxy): dedupe non-streaming passthrough guardrail handling
Replace the inline JSON/eventstream block in the streaming-request branch
with a call to _handle_non_streaming_allm_passthrough_route so both paths
share one implementation and cannot diverge.
* fix(bedrock): preserve trailing bytes when re-encoding converse stream
The event-stream re-encoder only emitted parsed frames, so any trailing
bytes left after the parse loop (truncated/corrupt final frame, or fewer
than 16 bytes after the last complete frame) were silently dropped from
the de-anonymized output. Capture and re-append them so the transformer
never truncates the stream.
* fix(proxy): guard non-dict post-call hook return on bedrock passthrough JSON path
* fix(proxy): guard malformed JSON body on bedrock passthrough guardrail path
* fix(proxy): close guardrail bypass via tool result text and default-mode post-call guardrails on bedrock passthrough
Pre-call extraction only read top-level Converse text blocks, so blocked
content placed under toolResult.content[].text was forwarded to Bedrock
without the key/team guardrail seeing it. Extraction now walks nested tool
result text and write-back mutates the owning block in place.
Post-call buffering for passthrough used _has_post_call_guardrails, which
excludes event_hook=None guardrails. Those guardrails run at post_call, so
their output processing was skipped and the raw upstream body was returned.
Add a passthrough-specific predicate that counts them.
* refactor(proxy): route bedrock event-stream de-anonymization through llm passthrough dispatcher
Remove the hardcoded bedrock provider guard from common_request_processing
by delegating event-stream de-anonymization to LlmPassthroughRouteHandler,
which resolves the provider from the existing handler registry. Keeps
proxy/ provider-agnostic and reuses the same dispatch path as the input
and output guardrail handlers.
Also log instead of silently dropping the result when post_call_success_hook
returns a non-dict on the JSON and event-stream passthrough paths.
* fix(proxy): close guardrail bypass on bedrock invoke passthrough routes
Pre-call extraction and post-call output processing only handled Converse
shapes, so /bedrock/model/{modelId}/invoke and invoke-with-response-stream
returned unguarded. An authenticated caller could move blocked content into
an InvokeModel payload and skip the key/team guardrail entirely.
Non-Converse Bedrock routes now fall back to the generic passthrough handler,
which scans the full request and response payloads so blocking guardrails
still run, matching how other passthrough providers are guarded.
* fix(proxy): keep non-bedrock passthrough streams streaming under post-call guardrails
* fix(bedrock): scan non-text converse blocks for passthrough guardrails
Key/team guardrails on bedrock converse passthrough only saw top-level
text blocks, so a caller could hide prompt content in toolUse.input or
toolResult.content[].json and have it forwarded to Bedrock without the
configured guardrail inspecting it, bypassing blocking guardrails by
default. Walk those arbitrary-JSON subtrees and write masked values back
in place. Extend the non-streaming converse response path to the
equivalent model-output fields (toolUse.input, reasoningContent text and
citationsContent text) while leaving structural values such as reasoning
signatures and citation sources untouched.
* fix(bedrock): make passthrough guardrail string collection iterative and type-safe
Rewrite _collect_strings with an explicit stack so it no longer recurses,
satisfying the recursive-function CI guard, and widen the holder container
type so mypy accepts indexing JSON nodes by str or int keys.
* fix(proxy): scope passthrough post-call guardrail buffering to the request
Buffering the Bedrock event stream into a single non-streaming response was
gated on whether any post_call guardrail existed globally, so every
converse-stream request lost streaming once any post_call guardrail was
registered, even for keys that did not reference it. Mirror the gate used by
post_call_success_hook (should_run_guardrail against the request's merged
guardrails) so only requests whose key/team actually trigger a post_call
guardrail are buffered.
* fix(bedrock): guardrail non-text converse stream deltas on passthrough
de_anonymize_event_stream only routed delta.text through the post-call guardrail, so model output streamed in reasoningContent.text, toolUse.input or citationsContent.content[].text was forwarded raw and skipped masking/blocking. Collect every user-visible text field per contentBlockDelta, concatenate per logical stream so split mask tokens still reassemble, run them through the hook, then redistribute the guardrailed text back into the matching delta fields. This brings streaming coverage in line with the non-streaming Converse output handler.
* refactor(proxy): keep bedrock event-stream content-type detection in llms
Move the vnd.amazon.eventstream content-type check out of the proxy
passthrough path into BedrockPassthroughGuardrailHandler via the
LlmPassthroughRouteHandler dispatcher, so proxy code stays
provider-agnostic. Also patch the actually-called
_has_post_call_guardrails_for_passthrough in the malformed-body
regression test instead of the unused _has_post_call_guardrails.
* fix(proxy): forward upstream headers on bedrock guardrail passthrough responses
Mirror the non-guardrail passthrough path by merging the upstream
response headers (via get_response_headers) into the guardrailed
non-streaming and event-stream responses, so headers like
x-amzn-requestid survive when a post-call guardrail rewrites the body.
Drop the stray fastapi HTTPException import from the SDK-tree handler
test in favor of a local sentinel exception.
* fix(bedrock): scan tool definitions and additional request fields for passthrough guardrails
Converse passthrough guardrails only scanned system and message content, so
a key holder could route blocked or PII text through toolConfig tool names,
descriptions and input schemas or through additionalModelRequestFields, all
of which are still forwarded to Bedrock. Collect strings from those fields
too so key/team guardrails inspect and rewrite them, matching how the
chat-completions path forwards tool definitions to guardrails.
* fix(bedrock): log instead of silently dropping passthrough guardrail edge cases
* test(local): skip httpbin timeout probe when the service returns 5xx
local_testing_part1 was failing on test_post_delay_exceeds_per_request_timeout_raises
because httpbin.org/delay/10 intermittently answers 503 instead of delaying, so
HTTPHandler.post raised MaskedHTTPStatusError rather than the expected Timeout. The
test already means to skip when httpbin is unavailable, but its guard only probed
GET /get and ignored a server error on the delay endpoint. Treat a 5xx from httpbin as
'service unavailable' and skip, which is outside this repo's control, while still
asserting Timeout when httpbin genuinely delays.
* fix(proxy): set content-type on buffered bedrock passthrough event-stream responses
* fix(bedrock): skip passthrough output write-back when guardrail returns no texts
* fix(proxy): apply response-headers hook on guardrailed bedrock passthrough responses
* refactor(bedrock): import event-stream crc32 from binascii not botocore internals
* fix(proxy): scope bedrock passthrough stream buffering to de-anonymizable endpoints
Only buffer a passthrough event stream into a non-streaming response when the
resolved provider and endpoint actually have an event-stream guardrail handler
that can rewrite frames (Bedrock converse-stream). Other Bedrock event-stream
endpoints such as invoke-with-response-stream keep streaming, since the Converse
handler leaves their frames untouched and buffering would silently break the
streaming contract for no content change.
* test(ui-e2e): re-issue deep-link navigation when auth bootstrap drops the page param
navigateToPage deep-links to /ui?page=<page> then proceeds once the network
settles, but a fresh load can race the auth bootstrap: the app momentarily
treats the session as anonymous, bounces through /ui/login, and returns to the
default Virtual Keys page with the ?page= query param dropped. The helper never
checked where it actually landed, so any single bounce left callers asserting
against the wrong page and timing out (mcpServers, modelHub, addModel).
Confirm the requested page is what rendered and re-issue the navigation when it
was clobbered; auth is warm by the second load so the param sticks. Migrated
path routes are left alone since they intentionally leave the legacy root.
---------
Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com>
|
||
|
Mateo Wang
|
20e453f698
|
feat(cli): per-agent lite claude / codex / opencode commands that wrap coding agents through the proxy (#29850)
* feat(cli): add `litellm-proxy run -- <agent>` to wrap coding agents through the proxy Wraps Claude Code, Codex, OpenCode, and any other coding agent so all of its LLM traffic routes through a LiteLLM proxy, with the agent-vault style of "just works" DX: one `run -- <agent>` command, auto SSO login when interactive, env-key "agent mode" for containers/CI, and a fail-fast key check against the proxy so bad credentials error immediately instead of deep inside the agent. The wrapped binary is detected by name to pick the right variables. Claude Code gets ANTHROPIC_BASE_URL (the bare proxy root, so it appends /v1/messages) and ANTHROPIC_AUTH_TOKEN, with any stray ANTHROPIC_API_KEY cleared so the proxy token wins. Codex and OpenCode get OPENAI_BASE_URL (proxy + /v1) and OPENAI_API_KEY. Unrecognized commands get both sets so they work either way. `litellm-proxy claude-code` remains as a shortcut for `run -- claude`. The core logic is split into dependency-injected helpers (agent_profile, build_agent_env, verify_proxy_key, run_agent) so env wiring, the preflight, and the launch handoff are unit-tested without monkeypatching, alongside CliRunner tests for auth resolution, agent mode, and auto-login. Mutation-tested the env profiles, preflight, and agent-mode branch to confirm the tests fail when the behavior is broken. https://claude.ai/code/session_0154VpLXW7mMvk5wfbgPRJa6 * Make each coding agent its own litellm-proxy command Replace the `run -- <agent>` interface and the `claude-code` shortcut with top-level commands generated per known agent, so launching is just `litellm-proxy claude`, `litellm-proxy codex`, or `litellm-proxy opencode`, with everything after the agent name forwarded straight to it. This drops the ceremony of `run --` and cuts typing. The `--model`/`--small-fast-model` wrapper flags are gone; pass the agent's own model flag instead, or export the model env vars (the wrapper preserves what you already have set), which keeps the surface minimal and avoids intercepting flags the agent owns. Rename the module to agents.py to match. * fix(cli): route `litellm-proxy codex` through the proxy via a custom provider Codex ignores OPENAI_BASE_URL (it always dials api.openai.com over the Responses WebSocket transport), so the OpenAI env profile alone left `litellm-proxy codex` talking to OpenAI directly instead of the proxy. Point Codex at the proxy with a custom provider passed as `-c` config overrides, and force the HTTP/SSE Responses transport with supports_websockets=false since the proxy does not speak the Responses WebSocket protocol. The provider reads its key from OPENAI_API_KEY, which the agent env already exports. The overrides are injected ahead of the user's args so they precede Codex's subcommand. Claude Code and OpenCode are unaffected; they honor the exported env vars. Adds regression tests for the per-agent launch args and the injection ordering. Co-authored-by: Mateo Wang <mateo-berri@users.noreply.github.com> * Rename litellm-proxy CLI command to lite The proxy management CLI was invoked as litellm-proxy, which is a lot to type for an everyday command. Rename the console script entry point to lite and update the in-CLI usage examples, help text, error messages and docs to match. * fix(sso): stop CLI auth success page from hanging on "Closing..." The CLI opens the SSO success page with webbrowser.open, so the tab is not script-opened and the browser refuses window.close(). The countdown would end on "Closing..." and the tab would sit there forever. Drop the countdown and just show "You can now close this window and return to your terminal." from the start, while still attempting window.close() once so the tab auto-closes in the rare case the browser allows it. Add a regression test asserting the manual-close instruction is always present and the misleading countdown/"Closing..." text is gone. * fix(cli): reattach controlling terminal after SSO login, keep litellm-proxy alias When the first `lite claude` has to log in via browser SSO, completing the login could leave stdin detached from the terminal, so a TUI agent like Claude Code would start in non-interactive mode and exit with "Input must be provided". The wrapper now reopens the controlling terminal onto stdin just before handoff when the session started interactively; piped or redirected input is detected up front and left alone, so agent-mode and non-interactive use are unchanged. Also keep the `litellm-proxy` console script as an alias for `lite` so existing scripts and CI that invoke `litellm-proxy` keep working; both names map to the same CLI. * feat(install): make the curl installer need only curl, not a pre-existing Python The installer now lets uv provision a managed Python 3.13 when no suitable interpreter is found, instead of aborting. The minimum is also bumped from 3.9 to 3.10 to match the package's requires-python (>=3.10), so a system Python 3.9 is no longer selected only for uv tool install to reject it. * feat(cli): add thin litellm[cli] install path (install-cli.sh + brew) for the lite CLI On a developer laptop the `lite` CLI only needs `lite login` and running coding agents through a proxy, but the sole install path was `litellm[proxy]`, which drags in the whole server tree (fastapi, uvicorn, boto3, polars, cryptography, litellm-enterprise). The CLI's heavy imports are all guarded, so it runs on the base SDK plus just rich, pyyaml and requests. Add a `cli` extra carrying exactly those three, a `scripts/install-cli.sh` curl one-liner that installs `litellm[cli]`, and a `BerriAI/homebrew-litellm` tap formula with a release runbook under `packaging/homebrew/`. The installer passes no `--python`, so uv honours litellm's requires-python and provisions a managed interpreter, skipping a too-old (3.9) or too-new (3.14+) system Python instead of failing to resolve. A pyproject thin-contract test asserts the `cli` extra keeps the deps the CLI imports and never leaks a server-only dependency from `proxy`, so the laptop install cannot silently re-bloat * fix(install): let uv pick the Python via --python-preference system Both installers detected a system Python with a floor-only check and forced it with `uv tool install --python <interp>`. On a host whose only Python is outside litellm's requires-python (a too-old 3.9 or, increasingly, a too-new 3.14) that forced an incompatible interpreter and the resolve failed. Drop the detection and pass `--python-preference system`: uv reuses a compatible system Python when present and downloads a managed one otherwise, always honouring requires-python * test(router): filter aiohttp unclosed-session gc noise in test_async_fallbacks test_async_fallbacks asserts the last three captured log records are the router's fallback messages. Under the litellm_router_testing job (pytest -k router -n 4) many router tests share the module-level in_memory_llm_clients_cache (max 200, ttl 3600s). Older cached OpenAI/Azure clients get evicted while their aiohttp ClientSession is still open, and when the gc reclaims them aiohttp emits "Unclosed client session"/"Unclosed connector" through the asyncio logger. Those records land in caplog mid-test and push the expected router logs out of the last-three window, so the assertion flips to failing non-deterministically. These warnings are async cleanup noise, not router debug logs, so filter them out exactly like the existing leaked-task warnings before asserting order. The assertion on the three router fallback messages is unchanged. --------- Co-authored-by: Cursor Agent <cursoragent@cursor.com> Co-authored-by: Mateo Wang <mateo-berri@users.noreply.github.com> Co-authored-by: Claude <noreply@anthropic.com> |
||
|
Sameer Kankute
|
cb041966bf
|
Litellm oss staging 040626 (#29671)
* fix(azure): apply api_version fallback chain to image edit URL
`AzureImageEditConfig.get_complete_url` only read `api_version` from
`litellm_params`. When callers configured it via `litellm.api_version`
or `AZURE_API_VERSION`, the constructed URL had no `?api-version=` and
Azure responded `404 Resource not found`.
Apply the same fallback chain the Azure chat path already uses in
`common_utils.py`:
litellm_params > litellm.api_version > AZURE_API_VERSION env >
litellm.AZURE_DEFAULT_API_VERSION
Adds 5 unit tests pinning each layer of the chain plus a regression
guard for `api_base` that already carries `?api-version=`.
* feat(mcp): core sampling and elicitation flow with security hardening
- Add sampling_handler.py: full MCP sampling/createMessage flow with
model selection (hint-based + priority-based), auth enforcement,
budget checks, route restriction gates, and tag policy pre-auth
- Add elicitation_handler.py: MCP elicitation/create relay with
downstream client capability detection
- Wire sampling/elicitation callbacks in mcp_server_manager.py
gated behind allow_sampling/allow_elicitation config flags
- Add allow_sampling/allow_elicitation fields to MCPServer type
- Fix session lock deadlock: skip lock for JSON-RPC response POSTs
(elicitation/sampling replies) with truncated-body heuristic
- Extend client.py with sampling_callback and elicitation_callback
- Security: RouteChecks gate, tag-budget bypass fix, x-forwarded-for
spoofing fix, Latin-1 header encoding guard
- Add 4 new test modules (model access, priority selection, request
builder, tool conversion) + update existing MCP tests
* fix(security): run pre-call guardrails before MCP sampling acompletion
Without this, an upstream MCP server with allow_sampling enabled could
send prompts that bypass every guardrail (content filtering, PII
redaction, prompt-injection detection) configured on /chat/completions.
- Call proxy_logging_obj.pre_call_hook(call_type='acompletion') before
llm_router.acompletion so guardrails fire for sampling sub-calls
- Add HTTPException to the re-raise list so guardrail rejections
propagate correctly instead of being swallowed as generic errors
* feat(bedrock_mantle): add Responses API support (/openai/v1/responses) (#29490)
* feat(bedrock_mantle): add Responses API transformation config
* test(bedrock_mantle): cover trailing-slash api_base normalization
* feat(bedrock_mantle): export BedrockMantleResponsesAPIConfig
* feat(bedrock_mantle): register gpt-5.x Responses config (gpt-oss unchanged)
* feat(bedrock_mantle): add gpt-5.5/gpt-5.4 Responses price-map entries
* refactor(bedrock_mantle): exclude gpt-oss instead of allow-listing gpt-5 for Responses routing
Frontier OpenAI models on Bedrock Mantle are Responses-only on /openai/v1/responses;
gpt-oss is the legacy family that also speaks chat-completions. Gate by excluding
gpt-oss (which keeps its chat-completions emulation) and defaulting everything else
to the native Responses config, so future frontier models (gpt-6, etc.) route
correctly without a code change. Verified against the live us-east-2 Mantle endpoint:
gpt-oss 400s on /openai/v1/responses while gpt-5.5 400s on both standard paths.
* test(bedrock_mantle): cover supports_native_websocket opt-out
Closes the one uncovered line flagged by codecov on the Responses config.
The assertion documents that Mantle Responses has no realtime/websocket
transport, so realtime routing must not attempt a socket it cannot serve.
* fix(bedrock_mantle): route file_search through emulation instead of forwarding to Mantle
BedrockMantleResponsesAPIConfig inherited supports_native_file_search()
-> True from OpenAIResponsesAPIConfig but never overrode it. Mantle has no
OpenAI vector stores, so a forwarded file_search tool is rejected with a
400 (verified upstream: Tool type 'file_search' is not supported). Opting
out, like the existing supports_native_websocket override, routes the tool
through LiteLLM's file_search emulation instead.
* fix(bedrock_mantle): only route openai.gpt frontier models to Responses
The previous gate excluded gpt-oss and routed every other model to the
native Responses config. But on Mantle only the OpenAI gpt frontier models
(gpt-5.x) are served on /openai/v1/responses; gpt-oss and the non-OpenAI
families (nvidia, mistral, google, zai, ...) are chat-completions only and
400 on that path. Allow-list the openai.gpt- family (excluding gpt-oss)
instead, so chat-only models fall through to the chat-completions emulation.
Verified against the live us-east-2 endpoint: nvidia.nemotron-nano-9b-v2
returns 400 on /openai/v1/responses and 200 on /v1/chat/completions.
* feat(custom_llm): allow streaming/astreaming to yield ModelResponseStream (#27580)
* fix(custom_llm): allow streaming/astreaming to yield ModelResponseStream directly
* fix(streaming): enhance ModelResponseStream handling for custom LLM providers
* fix(streaming): strip finish_reason from content chunks and ensure tool_calls are preserved
* fix(streaming): add type ignore for finish_reason assignment in CustomStreamWrapper
* fix(proxy): strip stack trace from HTTP 503 responses (CWE-209) (#28330)
* fix(proxy/cwe-209): strip Python traceback from HTTP 503 error responses
The /cache/ping endpoint included a full Python traceback in its 503 error
response body (inside the ProxyException message), leaking internal file
paths, line numbers, and call stacks to any caller. Two MCP route handlers
in proxy_server.py similarly interpolated str(e) into "Internal server
error" detail strings.
Fix: log the traceback server-side via verbose_proxy_logger.exception()
and omit it from the ProxyException payload / HTTPException detail returned
to clients. Tests updated to assert no "traceback" keyword or frame paths
appear in the 503 body, with a new dedicated regression test.
CWE-209: Generation of Error Message Containing Sensitive Information.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix(proxy/cwe-209): apply Greptile P2 fixes and add MCP exception-path tests
Greptile 4/5 review identified two remaining gaps and Codecov reported
0% coverage on the two MCP handler exception branches:
1. caching_routes.py — str(e) in "Service Unhealthy ({str(e)})" could
still leak Redis hostnames/IPs; replaced with static "Service Unhealthy".
HTTPException is now re-raised before the generic handler so the
"cache not initialized" 503 still reaches callers with its detail.
Removed the redundant str(e) arg from verbose_proxy_logger.exception()
(exception() already appends the traceback automatically).
2. tests — two new unit tests cover the exception paths in
dynamic_mcp_route and toolset_mcp_route that were previously at 0%:
- test_dynamic_mcp_route_unexpected_exception_returns_500_without_traceback
- test_toolset_mcp_route_unexpected_exception_returns_500_without_traceback
All 25 tests pass (9 caching + 16 MCP).
CWE-209: Generation of Error Message Containing Sensitive Information.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* test(caching_routes): restore precise assertion in test_cache_ping_no_cache_initialized
The assertion was weakened to `"Cache not initialized" in str(data)`, which
matches the raw string of the entire response dict and would pass even if the
error moved to an unexpected field or changed structure.
Restore a targeted check on the parsed response: assert the exact string in
the correct field `data["detail"]`, matching FastAPI's HTTPException
serialisation format {"detail": "<message>"}.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* test(caching_routes): restore precise assertion and add CWE-209 no-cache path test
The assertion in test_cache_ping_no_cache_initialized was weakened to
`"Cache not initialized" in str(data)`, which matched against the raw string
representation of the entire response dict. This would pass silently even if
the error message moved to an unexpected field or the structure changed.
Restore a targeted assertion on the parsed field:
assert data["detail"] == "Cache not initialized. litellm.cache is None"
matching FastAPI's HTTPException serialisation format exactly.
Add test_cache_ping_no_cache_does_not_expose_internals to show the code path
is still working correctly after the CWE-209 fix: verifies that the HTTPException
is re-raised as-is (no traceback, no source paths), and asserts the complete
response structure is exactly {"detail": "Cache not initialized. litellm.cache is None"}.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix(caching_routes): restore ProxyException envelope for null-cache 503
The except HTTPException: raise guard (added in the CWE-209 fix) caused
the null-cache HTTPException to escape as FastAPI's {"detail": "..."} shape
instead of the {"error": {...}} ProxyException envelope that callers expect.
Move the null-cache guard before the try block and raise ProxyException
directly so the response structure is consistent with all other /cache/ping
503s, and the except HTTPException: raise guard is only reachable by
unexpected downstream HTTPExceptions.
Update the two no-cache tests to assert the correct ProxyException envelope.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
---------
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
* Update utils.py (#26609)
* feat(pricing): add Snowflake Cortex REST API model pricing (#26612)
* feat(pricing): add Snowflake Cortex REST API model pricing
## Summary
Adds pricing and context window information for 20+ Snowflake Cortex REST API models to `model_prices_and_context_window.json`.
## What's included
- **7 Claude models** (sonnet-4-5, sonnet-4-6, 4-sonnet, 4-opus, haiku-4-5, 3-7-sonnet, 3-5-sonnet) — with prompt caching rates
- **4 OpenAI models** (gpt-4.1, gpt-5, gpt-5-mini, gpt-5-nano) — with prompt caching rates
- **5 Llama models** (3.1-8b, 3.1-70b, 3.1-405b, 3.3-70b, 4-maverick)
- **1 DeepSeek model** (deepseek-r1)
- **1 Mistral model** (mistral-large2)
- **1 Snowflake model** (snowflake-llama-3.3-70b)
- **2 Embedding models** (arctic-embed-l-v2.0, arctic-embed-m-v2.0)
Each entry includes `input_cost_per_token`, `output_cost_per_token`, `cache_read_input_token_cost` (where applicable), `max_input_tokens`, `max_output_tokens`, and capability flags (`supports_function_calling`, `supports_vision`, `supports_prompt_caching`, `supports_reasoning`).
## Pricing source
All prices are in USD per token, sourced from the official [Snowflake Service Consumption Table](https://www.snowflake.com/legal-files/CreditConsumptionTable.pdf) — Tables 6(b) (REST API with Prompt Caching) and 6(c) (REST API).
## Context
The existing `snowflake/` provider has zero model entries in the pricing JSON, which means LiteLLM cannot track costs for Snowflake Cortex calls. This PR fills that gap.
## Related
- Existing provider: `litellm/llms/snowflake/`
- Cortex REST API docs: https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-rest-api
* Update model_prices_and_context_window.json
Fix the JSON parsing error
* Update model_prices_and_context_window.json
Removed the duplicate entry
* fix(utils): copy extra_body before adding unknown params to prevent model config mutation (#29620)
Fixes #29615. In add_provider_specific_params_to_optional_params, the line:
extra_body = passed_params.pop("extra_body", None) or {}
returns the original dict reference when extra_body is non-empty (truthy).
Subsequent writes like extra_body[k] = passed_params[k] then mutate the
shared model config object held by the router, poisoning /model/info and
all subsequent requests for that deployment.
The or {} short-circuit creates a new dict only when extra_body is falsy
(None or {}), which is why the bug does not reproduce with extra_body: {}.
Fix: wrap in dict() so we always work on a fresh shallow copy.
* fix(vertex_ai): Bake tool_choice into Gemini CachedContent body to prevent silent drop (#29097)
* fix(vertex_ai): bake tool_choice into Gemini CachedContent body to prevent silent drop
* address greptile feedback on tool_choice cache test
* adds test that uses ToolConfig(functionCallingConfig=FunctionCallingConfig(mode=ANY)) instead of a dict literal, mirroring what map_tool_choice_values actually produce
* fix(gemini/veo): move image from parameters into instances[0] (#29501)
* fix(gemini/veo): move image from parameters into instances[0]
Veo's predictLongRunning schema puts image (and prompt) on the
instances element; parameters is for aspectRatio/durationSeconds/etc.
The Gemini path was leaving image in params_copy, so it ended up
nested under parameters and the API silently ignored it.
The Vertex path already builds the instance dict explicitly, so this
just aligns the Gemini path with it.
Fixes #29498
* address greptile: unconditional pop + BytesIO test
- Pop `image` from params_copy unconditionally so it never reaches
GeminiVideoGenerationParameters even when None, removing implicit
reliance on Pydantic's extra-field-ignore.
- Add test_transform_video_create_request_image_filelike_goes_to_instance
covering the BytesIO path (_convert_image_to_gemini_format) — round-trips
the base64 to confirm encoding.
- Add test_transform_video_create_request_image_none_is_dropped covering
the new None branch.
* fix(huggingface): handle special token text in embedding usage (#29660)
* fix(guardrails): recompile ToolPermissionGuardrail rules on update_in_memory_litellm_params (#29655)
* fix(guardrails): recompile ToolPermissionGuardrail rules on update_in_memory_litellm_params
ToolPermissionGuardrail builds self.rules and the compiled target/pattern
maps only in __init__. The base update_in_memory_litellm_params re-sets raw
attributes via setattr but never rebuilds those maps, so a guardrail updated
in place (PUT /guardrails, or the immediate in-memory sync) keeps enforcing
the construction-time rules until it is reinitialized (PATCH path, periodic
DB poll, or restart).
Extract the compile step into _load_rules and override
update_in_memory_litellm_params to rebuild from it (dict- and model-safe),
re-normalizing default_action / on_disallowed_action. Mirrors the existing
PresidioGuardrail override of the same method. Adds regression tests.
Fixes #29592.
* fix(guardrails): handle dict params in ToolPermissionGuardrail in-memory update
Delegate to super() only for LitellmParams input (the base setattr loop is
model-only); apply the raw-dict case inline. Fixes the mypy arg-type error
and makes the recompile work when the proxy passes the raw DB dict.
* fix(guardrails): preserve tool-permission rules on a partial in-memory update
A partial update (e.g. a LitellmParams whose rules field is None) ran through
the generic setattr, which set self.rules to None, and the recompile was
skipped, leaving the guardrail with no rules. Snapshot the previous rules and
restore them when the update carries no rules; an explicit empty list still
clears them. Adds a regression test for the rules-absent case.
Addresses the Greptile review note on #29655.
* fix(bedrock): stop base_model label from stripping tools/tool_choice (#29621)
* fix(bedrock): stop base_model label from stripping tools/tool_choice
A Router/proxy Bedrock deployment whose model_info.base_model is a friendly
label (e.g. claude-haiku-4-5) silently lost tools/tool_choice: the outgoing
Converse request was built without toolConfig, so the model behaved as if no
tools were provided. Worked in v1.84.0, regressed in v1.85.0, and with
drop_params=true it failed silently.
Two changes compound into the bug. completion() passed model_info.base_model
as the model argument to get_optional_params, so the real Bedrock model id
never reached supported-param resolution; and get_supported_openai_params
resolved the provider config's params from base_model or model, letting the
label fully replace the real model. For Bedrock the label resolves to no tool
support, so tools/tool_choice were dropped before transformation.
completion() now keeps model as the real deployment model and threads the
resolved base_model (kwarg or model_info) through separately, and
get_supported_openai_params treats base_model as additive: it returns the
union of the params supported by model and by base_model. A hint can only add
capabilities, never strip ones the real model already exposes, which also
preserves the original base_model behavior from #27717 and Azure's base_model
driven model-type detection.
Fixes #29618
* test(main): make base_model param test robust to new parametrize cases
Restore an explicit per-case expected_model_param literal instead of
hardcoding the gemini id, so a future case with a different model can't
produce a misleading assertion failure.
* fix(fireworks_ai): pass response_format json_schema through unchanged (#29606)
FireworksAIConfig.map_openai_params was rewriting the OpenAI strict
`{type: json_schema, json_schema: {name, strict, schema}}` shape into
`{type: json_object, schema: ...}` before sending to Fireworks, dropping
`strict` and `name` and changing the `type`. Per Fireworks' docs json_object
means "force any valid JSON output (no specific schema)", so the schema
constraint was effectively dropped and grammar-guided decoding never ran;
model output silently violated the schema.
The rewrite landed in #7085 (Dec 2024) when Fireworks did not yet accept
native json_schema. Fireworks accepts the OpenAI strict shape natively now,
so the rewrite has become a regression.
Removes the rewrite. Passes response_format through unchanged. Updates the
existing test_map_response_format to assert pass-through. Adds focused
regression tests in tests/test_litellm/ covering preservation of type,
strict, name, and schema body, plus that json_object alone still works.
* fix(types): import Required from typing_extensions in gemini types
* style: reformat sampling_handler.py for py312 black compat
* refactor(mcp-sampling): extract helpers to fix PLR0915 too-many-statements in handle_sampling_create_message
* fix(proxy-server): add explicit ProxyLogging type annotation to proxy_logging_obj to fix mypy inference
* fix(mcp-sampling): suppress mypy assignment error on ImportError fallback for proxy_logging_obj
* fix(test): use .value when comparing LlmProviders enum against string in test_default_api_base
* fix(test): iterate LlmProviders enum in test_default_api_base to avoid str pollution from custom provider registration
litellm.provider_list is a mutable global initialized to list(LlmProviders) but custom_llm_setup() appends plain provider strings to it. When a test_custom_llm.py test runs first in the same xdist worker, provider_list contains a str and calling .value on it raises AttributeError. Iterate the immutable LlmProviders enum instead, which is deterministic and what the check intends.
* fix(mcp): depth-aware JSON-RPC response detection and neutral speed-priority fallback
Replace the flat substring check in the truncated-body routing path with a
top-level-key scan so a JSON-RPC response whose result payload nests a
"method" field is still detected as a response and skips the session lock,
removing a deadlock against the in-flight tool call awaiting it.
Drop the inverse max_output_tokens speed proxy when no model exposes
output_tokens_per_second; context-window size does not track latency, so a
neutral score avoids biasing speedPriority toward the smallest-context model.
* fix(guardrails): make ToolPermission rule reload atomic on invalid regex
_load_rules appended each rule to self.rules before compiling its regex, so an
invalid pattern raised mid-loop after the bad rule was already live but without
a _compiled_rule_targets entry. _matches_regex reads a missing compiled target
as a None pattern and returns True, turning the bad rule into a match-all that
silently applies its decision to every tool. Via update_in_memory_litellm_params
(PUT /guardrails) this corrupted the live guardrail.
Build the parsed rules and compiled maps into locals and swap them in only after
every regex compiles, and restore the previous ruleset if a live update is
rejected, so an invalid regex now fails the update without leaving the guardrail
enforcing a broken policy.
* test(mcp): cover sampling conversion, model resolution, and elicitation relay paths
The MCP sampling and elicitation handlers shipped with partial test
coverage, leaving the response-to-MCP conversion, the model resolution
fallback chain, completion-kwargs assembly, guardrail routing, and the
entire elicitation relay untested. That pulled the PR's diff (patch)
coverage below the codecov threshold even though overall project
coverage rose.
Add focused unit tests for _convert_openai_response_to_mcp_result,
_convert_mcp_tools_to_openai, _convert_mcp_tool_choice_to_openai, image
and audio content conversion, the hint-matching and fallback branches of
_resolve_model_from_preferences, _build_completion_kwargs, the router and
guardrail-rejection paths of _run_guardrails_and_call_llm, the
handle_sampling_create_message success and error-propagation flows, the
marker-hoisting fallback for tool content on unexpected roles, and the
elicitation form/url/generic relay together with its decline paths
---------
Co-authored-by: shin-berri <shin-laptop@berri.ai>
Co-authored-by: yuneng-jiang <yuneng@berri.ai>
Co-authored-by: lengkejun <lengkejun@xd.com>
Co-authored-by: Yug <yugborana000@gmail.com>
Co-authored-by: Kent <72616338+kingdoooo@users.noreply.github.com>
Co-authored-by: tanmay958 <53569547+tanmay958@users.noreply.github.com>
Co-authored-by: DrishnaTrivedi <142084770+DrishnaTrivedi@users.noreply.github.com>
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-authored-by: Navnit Shukla <Navnit.shukla25@gmail.com>
Co-authored-by: PRABHU KIRAN VANDRANKI <72809214+VANDRANKI@users.noreply.github.com>
Co-authored-by: Adrian Lopez <109683617+adriangomez24@users.noreply.github.com>
Co-authored-by: hcl <chenglunhu@gmail.com>
Co-authored-by: JooHo Lee <96564470+BWAAEEEK@users.noreply.github.com>
Co-authored-by: Dinesh Girbide <85330597+Dinesh-Girbide@users.noreply.github.com>
Co-authored-by: cloudwiz <22098246+andrey-dubnik@users.noreply.github.com>
Co-authored-by: Ahmad Khan <ahmadkhan2508@gmail.com>
Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com>
|
||
|
Mateo Wang
|
b4aee2c7dd
|
test(vcr): close out the remaining VCR live-call leaks (#29603)
* Fix remaining VCR live-call leaks * test(vcr): dedupe live-test helpers and drop spurious kwargs Extract the duplicated isVertexQuotaError/runVertexRequestOrSkip Vertex quota-skip helpers into tests/pass_through_tests/vertex_test_helpers.js and the duplicated _skip_live_prompt_caching_test guard into tests/_live_test_helpers.py so each lives in one place. In test_aarun_thread_litellm, build a separate message_data carrying role/content for add_message and a thread_data without them for run_thread/run_thread_stream/get_messages, which no longer receive the spurious message fields. * test(overhead): assert mock transport is exercised in non-streaming and stream tests |
||
|
Mateo Wang
|
3f33efdd57
|
fix(tests): drop import-time completion call in test_register_model (#29521)
* fix(tests): drop import-time completion call in test_register_model test_update_model_cost_via_completion() was invoked at module scope, so it ran during pytest collection and fired a live OpenAI completion. The local test jobs glob the whole tests/local_testing folder and let pytest import every file, narrowing what runs only afterward with -k, so this call executed in every one of those jobs regardless of their filter. When the request failed (for instance a 429 once the OpenAI account hit its quota), collection of the file errored and aborted the entire session, which is why langfuse, assistants, router and local_testing_part2 all reported "ERROR collecting tests/local_testing/test_register_model.py" and never ran their own tests. Remove the stray call and add a regression that parses the module and fails if any locally defined function is invoked at module scope again * test: also guard async def from module-scope invocation ast.AsyncFunctionDef is a distinct node from ast.FunctionDef, so an async test invoked at module scope would have slipped past the guard. Collect both kinds of definitions * fix(responses): send Content-Type application/json on OpenAI responses requests OpenAI's responses API now rejects body-less requests (GET/DELETE) that arrive without a content type, returning 500 "Unsupported content type: 'application/octet-stream'. This API method only accepts 'application/json' requests". litellm's create path got the header for free because httpx sets it when a json body is present, but the delete/get handlers send no body and so sent no content type. The official OpenAI SDK declares Content-Type: application/json on every request; mirror that in validate_environment so all OpenAI responses calls carry it. This is what made tests/openai_endpoints_tests/test_e2e_openai_responses_api.py::test_basic_response fail on the responses.delete() call. |