mirror of
https://github.com/BerriAI/litellm.git
synced 2026-11-11 08:38:24 +00:00
97 commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
yujonglee
|
11af2eb22d
|
refactor(rust_bridge): route every Rust-to-Python call through host modules (#50822)
* refactor(rust_bridge): route every Rust-to-Python call through host modules Rust now reaches Python only through litellm/rust_bridge/host/ (plus the execution driver), and no Python module outside host/ imports it. Library calls that reproduce Python behavior (httpx and OSError error text, huggingface_hub downloads, the random generator, os.environ) go through host functions too. The rule is written into host/AGENTS.md The live litellm.model_cost table, its version and every Python writer move to litellm_core_utils/model_cost_table.py, and host/model_cost.py keeps only the functions Rust calls. The cost map fetch passes use_http_settings instead of handing Rust an http_settings() snapshot * refactor(rust_bridge): read REDIS_USERNAME from the process environment in Rust The Redis bridge asked Python for all of os.environ to read one key. CPython's os.environ writes through to the process environment, so Rust reads it with ProcessEnvironment like the rest of the module. Also drops doc comments that restated their code. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * ci(rust): raise the native extension size cap to 55 MB Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
yuneng-jiang
|
906578df39
|
test(integration): recognize the model list's lock in the held write helper (#50974)
The registry lock fix (#50933) makes add_deployment and set_model_list wait for a held delete on the same lock, taken inside HostedModelList as self._lock. The helper only knew lines spelled with self._deployment_index_lock, so it never saw those two writes reach the lock and timed out after 10 seconds. |
||
|
devin-ai-integration[bot]
|
d47c03175b
|
fix(responses): keep unset-effort gpt-5.4/5.5 tool calls on chat completions (#50968)
* fix(responses): keep unset-effort gpt-5.4/5.5 tool calls on chat completions Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * test(responses): explicitly route bridge fixtures to responses Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> --------- Co-authored-by: kenneth <kenneth@berri.ai> Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> |
||
|
devin-ai-integration[bot]
|
5299914964
|
test(http2): drain client traffic after sending GOAWAY (#50972)
Co-authored-by: mateo <277851410+mateo-berri@users.noreply.github.com> Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> |
||
|
yuneng-jiang
|
433163a8e2
|
test(integration): empty the shared client cache before the tier pin batch test (#50854)
The test stores its recording Redis client in litellm's shared client cache with the default ttl. When earlier tests in the same process have filled that cache with longer lived clients, the next insert evicts the recording client, the pin is read through a fresh client, and the test sees no commands on the wire. |
||
|
yujonglee
|
523b3efea2
|
refactor(model-cost): move the cost map inputs, fetch and register_model overlay into Rust (#50501)
* refactor(model-cost): move the cost map inputs, fetch and register_model overlay into Rust * fix(model-cost): keep main's compat helpers pure and move network tests to integration ModelCostMapReloaded.model_cost_map now returns the alias-sharing table, so an alias registration after adopting a reload still reprices the owner. _finalize_model_cost_map reads a caller's map without size limits and installs nothing, through the new NativeModelCostMap.finalize. The unit tests that started a local server or a child interpreter now live in tests/integration/sdk/test_model_cost_fetch_sdk.py, and run_isolated gets direct tests for joining its blocking pool and leaving the fork gate alone. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
yuneng-jiang
|
71c1a2aff6
|
test(integration): repair tests that merged changes left stale (#50782)
Several integration tests on main still assert behavior that later changes replaced on purpose, so they fail on every scheduled run, and one of them stops the whole sdk group at import. The batch cost case expected cached tokens to bill at the Batch input price when a price row has no Batch cache price. Since #50062 they bill at the Standard cache price, so the case now expects 0.000235 input cost and is renamed to say so. The OTel v2 attribute test pinned langfuse_otel and langtrace as carrying no request metadata baggage. Since #50373 every OTel v2 destination carries it, so all seven callbacks share one expectation and the Langfuse baggage test that was deferred under LIT-9128 runs again. The clean-EOF outage test trips its database relay on the text of the spend log insert. The ledger writes that insert without the "public" schema prefix, so the relay never tripped and the test timed out before it checked anything. The trigger now matches the insert on both write paths. The usage routing tests expected cooldown, tpm and rpm keys in one MGET. Since #50170 the cooldown keys are read in their own MGET, at most once per DEFAULT_COOLDOWN_REDIS_READ_INTERVAL_SECONDS. Both the proxy and the SDK test now expect the two MGETs, and the proxy test waits out that interval after its /model/info poll, which reads the same keys. The concurrent deployment delete test held its writer inside Router._update_deployment_indices_after_removal, which went away with the Python router core in #50178, so the file failed at import and stopped the whole sdk group. It now holds the writer where the registry reads the model list again after router-core applied the change. |
||
|
Mateo Wang
|
4679b9f642
|
test(copilot): align login guards with direct access tokens (#50285)
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> |
||
|
Nate Armstrong
|
300db9f01d
|
test(router): expect no fallback errors header for non-bool include_fallback_errors (#50259)
The Rust core reports fallback errors only when include_fallback_errors is literally true. The routing and SDK wire tests still expected the header for truthy non-bool values (1 and a 5 KB string), which the deleted Python core produced by accident. Every non-bool value now expects no errors, and the flag still must not reach the provider. |
||
|
quietjules
|
5a392b9e5f
|
fix(bedrock)!: keep Claude's thinking when response_format json_schema is set (#49696)
* fix(anthropic): stop forcing the response_format json tool when Claude will think A forced tool call skips Claude's thinking. The json_schema emulation forced json_tool_call unless thinking.type was enabled, so adaptive thinking and the models that think by default (Opus 5, Sonnet 5, Haiku 5.5, and the always-on families) lost their thinking on every structured-output request. AnthropicModelInfo.response_format_tool_can_be_forced is now the one check the Anthropic, Bedrock Converse (including the native output fallback) and Databricks paths use. It leaves the tool unforced when thinking is requested in any form or the model thinks by default, and keeps forcing it when the request turns thinking off (thinking disabled or reasoning_effort none) on a model that accepts forced tools. The new thinking_on_by_default cost map key marks Opus 5, Sonnet 5 and Haiku 5.5 on the providers that reach these paths, and a miss-only fallback rule covers unmapped Claude 5 and later ids. * fix(bedrock): route native output and thinking token counts by whether Claude will think Bedrock's native output opt-in now uses the same will-think check as the json tool, so Claude 5 models that think by default with no thinking param go native. Converse also asks for /usage/output_tokens_details on those models, so reasoning tokens are reported for them. * fix(anthropic): tell Claude to answer through the unforced response_format tool Moves the integration tests that check the forced JSON tool from Claude 5 models to Opus 4.8, since Claude 5 models now keep their default thinking and take the unforced tool or native output * fix(anthropic): keep open-object schemas on the json tool when Claude will think Native output closes every object, so an open object came back empty. Also declare thinking_on_by_default in the Rust model catalog and expect the Converse thinking-token field in test_main * fix(anthropic): hold streamed prose until the JSON tool decides the answer In json mode with the unforced JSON tool, Claude can write its working before it calls the tool. The stream now holds that text, drops it when the JSON tool answers, and sends it when Claude calls a caller tool or skips the JSON tool, matching the non-streamed reply. * test(anthropic): build the stream test events without nested comprehensions * fix(anthropic): keep always-on thinking models unforced, open objects native when thinking, and the tool instruction Claude-only * test(anthropic): keep main's open-object integration cases and cover the unforced tool on a default-thinking model * fix(anthropic): treat thinking False as thinking off so the JSON tool stays forced Cover thinking False on the Anthropic and Converse forcing tests, and pin that a Converse json_schema stream with the unforced JSON tool never sends Claude's working text as content * fix(anthropic): drop the thinking False forcing case that completion never reaches litellm.completion turns thinking False into no thinking param before any route sees it, so the case was dead on the request path and False keeps meaning the model's default * test(bedrock): collect the Converse json_schema stream chunks in one assignment --------- Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> |
||
|
Emerson Gomes
|
ecb8ef5ca8
|
feat(realtime)!: support latest OpenAI audio models (#35600)
* feat(realtime): support latest OpenAI audio models * fix(transcription): type SDK stream duration metadata explicitly * fix(realtime): preserve transport settings and translation duration * refactor(realtime): reuse SDK client initialization * fix(audio): bill finalized streams and authorize transcription models * refactor(audio): keep transcription pricing selection together * fix(realtime): reconcile model metadata and test contracts * fix(realtime): preserve provider duration and model capabilities * fix(realtime): keep transcription validation compatible with cost map guard * fix(realtime): address transcription and billing review findings * fix(realtime): preserve input usage and provider routing * fix(realtime): bill retained output and route Azure aliases * fix(realtime): reject malformed provider durations * fix(realtime): preserve batch cache rates and Azure v1 routes * fix(transcription): reject nonfinite provider durations * refactor(realtime): align annotations with current lint rules * fix(realtime): preserve translation deployment pricing and input usage * test(transcription): verify synchronous stream finalization * fix(realtime): retain missing translation input usage at close * fix(transcription): reject streams that bypass output guardrails * fix(realtime): reconcile audio protocols billing and current CI * fix(realtime): trust confirmed formats and authorize default models * fix(realtime): preserve absent models and reject unsupported guardrails * fix(realtime): reject translation policy pipelines * chore(ci): rerun transient infrastructure timing failure * fix(realtime): reconcile audio protocols with current upstream * fix(audio): validate provider responses at SDK boundaries * fix(realtime): preserve minimum OpenAI SDK compatibility * fix(realtime): preserve deferred Live session authorization * fix(realtime): preserve upstream authentication and transcription behavior * fix(audio): scope provider duration precedence to configured models * fix(typing): avoid Final declarations inside transcription cost loop * refactor(audio): reuse transcription mapping and duration resolution * fix(realtime): enforce model budgets and preserve translation routing * fix(audio): log streamed transcripts whose done event reports detected languages * fix(audio): pass stream, languages and keywords through unchecked for non-OpenAI transcription providers Providers without a streaming transcription handler got UnsupportedParamsError for stream=True, languages and keywords, where the merge base forwarded them as provider-specific params. Only openai and azure validate them now. * fix(audio): bill streamed gpt-transcribe from the provider's reported duration The stream path computed cost after logging had replaced the transcription's duration usage, so it fell back to the measured file length while the unstreamed call billed the provider's seconds. * fix(transcription): log streamed transcription usage from the provider response The stream logging path built the spend payload from the response after its duration usage was transformed away, so whisper-1 with stream=true and streamed gpt-transcribe logged an empty usage object. Bill and log from the provider's TranscriptionResponse in both the sync and async paths. * fix(realtime): keep main's routing, protocol and transport behavior around the new translation routes - Send OpenAI realtime and translation client_secrets and calls through litellm's httpx client again, so ssl_verify and custom CA settings apply and upstream errors keep main's shape - Use the input transcription model as the session model only for transcription sessions - Drop the new /openai/v1 translation aliases and record each lazy feature's own routes, so the /openai pass-through and the MCP routes resolve as on main - Let a configured Azure realtime_protocol win for every model, with GA as the default only when none is set - Return x-litellm headers on streamed transcriptions and mark the request done - Refuse translation sessions before connecting upstream when realtime guardrails apply or the provider is not OpenAI or Azure - Accept json and text for gpt-transcribe, and drop other formats under drop_params instead of failing - Bill translation sessions on output audio seconds only, at the single published rate - Document the 201 SDP answer on the translation calls routes and the transcription 200 response * fix(realtime): restore main behavior for guardrails, Azure transcription, and realtime clients Streamed transcription runs unstreamed when a deployment post_call guardrail applies, so the guardrail still sees the transcript. Azure transcription maps v1, latest and preview to the default dated version on the deployments route for every model. The SDK client secret keeps session.model ahead of model, realtime WebSockets connect to api_base again, realtime metadata keeps only guardrails, and managed agents need an explicit transcription model. Reverts the two unrelated streamed Responses error codes and the duplicate cached_tokens_details field, and the gpt-transcribe format check only reads the request provider's cost map entry. * fix(realtime): keep main's transcription, client secret and route behavior where the PR changed it Transcription reaches every provider as on main again, sends stream only when asked, forwards language and languages together, keeps form key order, bills provider seconds only when above zero, honors base_model for the new flags, keeps main's Azure api_version on the proxy path, and streams carry chat's headers, an error event on a mid-stream failure, and a spend row when the stream ends early. Translation cost reads both per-second rates from one source and spend rows carry the billed seconds. Realtime transcription WebSockets keep model in the upstream URL outside OpenAI's hosts, client secrets forward explicit nulls, accept conversation and delay, and answer 400 on a malformed body, the /openai translation paths are back to main, translation refuses only for guardrails realtime sessions run, nested aliases resolve and later session.update models are authorized, translation client secrets run per-model auth, beta clients bridged to GA get beta-shaped session events, and transcription sessions send the deployment model in both shapes. * fix(realtime): address review findings on transcription billing, streaming and realtime auth Per-second pricing replaces token billing only when the priced entry sets no token rates, translation rates fall back per key to the cost map, transcription cache keys include languages and keywords, a stream request a post_call guardrail forces unstreamed comes back as redacted SSE, Azure beta clients are bridged only for GA-only deployments, nested transcription models on the realtime WebSocket are checked against the key, and translation spend carries key and team metadata * fix(azure): pick the GA realtime protocol for deployments whose model_info.base_model is GA-only * fix(realtime): preserve prior merge resolutions after rebase * fix(realtime): enforce translation policies and normalize Azure transcription URLs * fix(cookbook): stop translation once output idles after the upload ends * fix(realtime): bill translation on input audio and drop pre-read stream cost headers Translation spend now uses the input audio seconds the session sent, priced through input_cost_per_second, instead of the translated output duration. A streamed transcription no longer returns cost headers before the stream is read, since the cost is only known after the last event. Also removes an unused realtime guardrail helper. * fix(realtime): drop top-level session nulls, bill dropped transcription streams, and bridge beta response fields Client secret, translation and transcription session bodies drop top-level nulls again, as main did, while nested nulls such as turn_detection still pass through. Transcription client secrets send the deployment model as session.model. A streamed transcription the client drops still runs its success callbacks, and the stream ends with data: [DONE] like OpenAI's. The beta to GA bridge maps max_response_output_tokens, speed and noise reduction, remaps response.create, and drops temperature under drop_params. The Entra token for realtime HTTP routes is fetched off the event loop * test(e2e): cover realtime translation, live transcription, client secrets, and streamed transcription * fix(realtime): close review gaps in translation billing, transcription streams, and Azure GA Bill translation sessions from the deployment's own per-second rates, add per-second prices for Azure gpt-4o transcribe models, keep stream-scoped guardrails seeing the caller's stream flag, map mid-stream transcription errors to LiteLLM exceptions with a [DONE] frame, map GA responses back to the beta shape for bridged beta clients, honor the minting key's guardrail opt-out on translation calls, restore the Azure transcription session URL, and flag GA-only Azure realtime models * fix(realtime): bill a translation rate the deployment omits from the session model * test(e2e): check translated speech is not English instead of exact French wording * fix(realtime): keep main's behavior for non-bridged backends, stream error frames, gateway URLs and Azure transcribe prices Remap beta events to GA only for bridged beta clients and translation sessions, bridge beta clients on GA-pinned Azure deployments, log error-only upstream sessions as failures, flag dated Azure GA realtime models, drop the unpublished Azure per-second transcribe rates, log a mid-stream transcription failure once, keep main's SSE error frame outside transcription, drop unsupported transcription params under drop_params, keep the caller's Azure auth headers, keep a gateway /v1 path, let key opt-outs release string-callback guardrails on translation, and keep the cookbook waiting while a long upload is still sending * fix(realtime): keep the proxy's realtime session gate off translation sessions A key that opts out of a legacy callback guardrail now gets a working translation session, since the transcription-guardrail gate no longer sends a realtime session.update the translation upstream rejects * fix(realtime): keep websocket model limits, bill measured transcription length, route transcription sessions on the allowed model A WebSocket opened without a query model now authorizes as an empty model, transcription cost prefers the measured upload length over the provider's reported seconds, transcription sessions overwrite nested models with the authorized top-level model again, and a non-boolean transcription stream value answers 400 or is dropped under drop_params. * fix(ci): copy the rust bridge extension into the openapi docs gate's base worktree * fix(model-catalog): accept the realtime translation endpoints in the cost map * fix(cookbook): close the translation session and read until session.closed * fix(realtime): treat an empty transcription stream as unset and pass the cost-map guard An empty `stream` form value runs the transcription unstreamed again, as OpenAI does, instead of answering 400. The four translate cost-map entries drop the translation routes from supported_endpoints until the main branch's catalog parser accepts them, and the nested transcription model lookup uses the router registry entry point. * test(proxy): drop the mid-stream error frame pin that main's error shape change replaced * fix(transcription): bill uploaded seconds when a stream fails mid-transcript * fix(proxy): bill uploaded seconds when a streamed transcription breaks mid-stream * fix(audio): keep transcription's positional arguments and redact streamed error details - Move the new languages, keywords and stream options to the end of transcription's parameters, so a prompt or format passed by position keeps its meaning - Redact internal hostnames, paths and tracebacks from a streamed transcription's mid-stream error event, as chat streams already do - Move the CA-bundle check for the realtime HTTP routes, which serves real TLS on loopback, from the unit suite to the SDK integration suite * fix(audio): stream the guarded transcript to SDK callers that asked for a stream * fix(logging): keep a streamed transcript out of callbacks that turned off message logging A streamed transcription's assembled response reached success callbacks with its text intact under turn_off_message_logging or a callback's message_logging=False, because the shared streaming redactor had no transcription branch. It now redacts the text, as the unstreamed call already does * fix(proxy): stream a transcription the deployment streamed when the client sent no stream field --------- Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> |
||
|
yuneng-jiang
|
f8f12a270a
|
test(integration): repair stale contract fixtures (#49700) | ||
|
Mateo Wang
|
8c46b2c0d6
|
test(router): trace the deployment index lock in the file that owns it (#49520)
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> |
||
|
Nate Armstrong
|
87d91d78b7
|
refactor(router): route the router family through RoutingContext and close the outside surface (#49118)
Strategies, pre-call checks, cooldown and fallback helpers, and the auto, complexity and quality sub-routers now take a RoutingContext instead of the Router, and read the router only through registry(), settings(), health() and the context's forwarding methods. Router-free helpers move from router_utils to litellm_core_utils, outside callers of router state switch to entry points, and the surface allowlist shrinks to the endpoint methods plus the entry points |
||
|
devin-ai-integration[bot]
|
c213234783
|
feat(vertex_ai)!: use HTTP/2 httpx client for search_api vector store (#40631)
* feat(vertex_ai): use HTTP/2 httpx client for search_api vector store Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * test(httpx): exercise configured request path Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * test(httpx): verify HTTP/2 request negotiation Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * fix(httpx): preserve HTTP/2 for sync proxy mounts Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * build(deps): add h2 to proxy extra so vertex search HTTP/2 is on by default Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * fix(http_handler): isolate ssl context for http2 clients Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * fix(http): add an HTTP/2 opt-out and harden the Vertex AI Search HTTP/2 client litellm.http2 is now unset by default, so http2: false or LITELLM_HTTP2=False turns the provider's HTTP/2 default off. A caller's own SSLContext keeps the search on HTTP/1.1, the HTTP/2 async client honors DISABLE_AIOHTTP_TRUST_ENV, forced IPv4 transports get the CA bundle and client cert, and vector store search retries once when the server drops the connection with a GOAWAY. * fix(http): retry vector store search only on HTTP/2 GOAWAY and keep aiohttp pool and proxy behavior Scope the one retry to a connection Google closes with an HTTP/2 GOAWAY and log it. When HTTP/2 replaces the aiohttp transport, keep aiohttp's idle keepalive and connection cap, and only then skip env proxies under DISABLE_AIOHTTP_TRUST_ENV, so the global LITELLM_HTTP2 toggle keeps HTTPS_PROXY * test(http): cover the vector store search GOAWAY retry in unit tests and drop duplicate SSL cache comments * fix(http): open another HTTP/2 connection once 100 streams are in use httpcore keeps every request to an origin on one HTTP/2 connection and caps it at 100 concurrent streams, so request 101 and later waited with no timeout. The handlers now spread requests over extra connections, opening one only when every open connection has 100 streams in flight. The force_ipv4 wire tests now cover ssl_verify=False and the client certificate. * fix(vertex_ai): reuse one Vertex search OAuth token across searches Every vector store search built a new search config, so the token cache it carries was thrown away and each search fetched a new Google OAuth token with a blocking call on the event loop. Under a burst of searches that blocking fetch held back every HTTP/2 response. The provider config is now shared, so searches with the same credentials reuse one token until it expires. * fix(http): keep the HTTP/2 connection spread within the connection cap * fix(http): keep sync vertex search on HTTP/1.1, since httpcore's sync HTTP/2 is not thread-safe httpcore 1.0.9 hands two threads the same HTTP/2 stream id, so threaded sync searches failed with protocol errors. Only the async client, which the proxy uses, now defaults to HTTP/2, and the sync connection spread is removed. * fix(http): split the connection cap across HTTP/2 lanes Each lane had its own pool with the full connection cap, so an origin answering HTTP/1.1, or several origins on one client, could exceed max_connections once more than 100 requests were in flight. Lanes now open only while lanes times 100 streams stay under the cap, and each lane pool gets at most 100 of it. * test(http2): scope wire-test peer logs to each fixture instance * fix(http2): keep TE, force_ipv4 trust, evicted-client and opt-out behavior on the HTTP/2 path - drop TE values other than trailers on https requests through the HTTP/2 spread, since h2 refuses them - force_ipv4 transports keep httpx's default trust store (SSL_CERT_DIR) unless SSL is configured - the evicted client closer asks the HTTP/2 spread whether a request is in flight - LITELLM_HTTP2 set to 0, no or off, and a quoted "false" http2 setting, opt out of HTTP/2 - the wire test's held-search barrier waits on an event instead of polling * fix(http2): count proxy-mounted transports when checking an evicted client for in-flight requests * test(http2): move the HTTP/2 eviction cases that open a loopback server into the integration wire tests * fix(http2): read an unquoted http2: 0 as off and free the stream slot of a response dropped without closing * fix(http2): free the stream slot of a partly read response dropped without closing * test(http2): check that the evicted client and its stream close after a partly read response is dropped --------- Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> Co-authored-by: mrinal <mrinal@berri.ai> Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> |
||
|
tavren27
|
e7d3b24d7c
|
fix(logging): mark mirrored silent_model calls with is_silent_experiment in spend logs (#48888)
* fix(logging): mark mirrored silent_model calls with is_silent_experiment in spend logs and logging payloads * fix(router): keep the mirror flag when litellm_metadata is set and strip a client-sent is_silent_experiment Spend rows read router metadata from litellm_metadata when a call carries it, so the router now sets the flag there too. The proxy drops a client-sent is_silent_experiment with the other router-owned keys, so real traffic cannot be recorded as mirrored. * fix(logging): mark mirrored calls with a router-only key so clients cannot set is_silent_experiment Spend rows and the standard logging payload now derive is_silent_experiment from _litellm_silent_experiment, which only the router writes and the proxy and pass-through routes strip from client metadata. A caller's own is_silent_experiment key stays untouched, so it keeps its current effect on mirroring and on provider metadata, and unmirrored calls log false in both places. * test(langfuse): expect is_silent_experiment false on every logged generation * fix(logging): mark mirrored calls with a router-only enum value so no client route can set is_silent_experiment The router stamps SilentExperimentMarker.MIRRORED, which client JSON can never produce, so batches, Responses WebSocket frames and pass-through need no strip. Langfuse layers the flag after its requester-metadata dedupe. * fix(logging): set is_silent_experiment only on mirrored calls so real calls keep their payloads Real calls no longer gain an is_silent_experiment key in spend rows or the standard logging payload, so OTEL, S3, GCS and every other logger receive what they did before. The Langfuse regression test now runs the real prompt helper. * fix(logging): keep the mirror marker out of callback metadata and read the spend-row flag from the logging payload Logging now strips the router's mirror marker from the request metadata buckets before any callback or vendor sees it, without touching the router's own dicts, so retries keep it. Spend rows read is_silent_experiment from the standard logging payload, which survives the collector's JSON round trip, so collector-mode rows keep the flag too. * fix(logging): mark mirrored calls through a context variable so router usage counts them once * fix(router): leave callback-started calls unmarked and stop client flags from turning mirroring off Logging handlers now run outside the mirrored-call context, so an LLM call a success or failure callback starts while logging a mirrored call is not flagged as mirrored. The guard against mirroring a mirror reads the same context variable instead of request metadata, so a client-sent is_silent_experiment no longer skips the mirrors. The unused module-level get_standard_logging_metadata stops copying resolved keys from raw metadata. * fix(router): keep main's client flag mirroring skip and the other logging metadata keys A client's is_silent_experiment: true skips mirroring again as on main, the module-level get_standard_logging_metadata drops only the flag, and the silent experiment tests wait on a logged-call signal instead of polling a deadline * fix(logging): take the mirror flag in Prometheus labels and Datadog cost tags only from the router * test: type the new parametrized tests and keep the recorder from changing its argument * fix(datadog): keep client request tags from setting the silent experiment cost tag A request tag is_silent_experiment:true set the Datadog cost tag on a direct call. Only the router flag sets it now. Also covers the sync failure handler's mirrored-call reset with a failure function callback test. --------- Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> |
||
|
simonp9
|
dff474e450
|
fix(responses): replay the original Responses item order, phase, and boundaries on the next chat turn (#48912)
* fix(responses): replay the original Responses item order, phase, and boundaries on the next chat turn * fix(responses): keep tool-call-only stream content empty and drop unreachable replay guards * fix(responses): reject an assistant phase OpenAI does not define unless drop_params is set * fix(responses): keep other providers' reasoning and stream output unchanged and fall back on records that do not fit * fix(responses): label a structured output turn with its returned message's phase and keep lone surrogates from failing the checksum * fix(responses): leave a turn without a phase when its content message cannot be read * test(responses): add black box tests for chat turn history replay * test(responses): expect the replay record on bridged chat replies * test(responses): expect the replay record in bridged chat translation cases * fix(streaming): keep the served service tier on the finish chunk the stream adds * test(responses): read the expected chat message without JSON validation so the replay matcher stays --------- Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> |
||
|
Kerry Lu
|
e4932a4539
|
test(router): set string fallback targets after Router init in the sdk cooldown cases (#48954)
Co-authored-by: devin-ai-integration[bot] <158243242+devin-ai-integration[bot]@users.noreply.github.com> |
||
|
Kerry Lu
|
84c4c5aa5c
|
fix(router): cool down failing deployments when fallbacks use string targets (LIT-10238) (#48913)
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> |
||
|
Kerry Lu
|
e21de0e418
|
fix(logging): bill each SDK call its own native Bedrock guardrail trace when calls share one metadata dict (LIT-10239) (#48923)
Co-authored-by: devin-ai-integration[bot] <158243242+devin-ai-integration[bot]@users.noreply.github.com> |
||
|
moe-berri
|
3d7b94b34f
|
fix(ollama): forward embedding credentials, headers and timeout (#48707)
* test(integration): cover ollama embedding key, header and timeout forwarding * fix(ollama): forward api_key, headers and timeout on native embeddings Co-authored-by: Mrun25 <165297479+Mrun25@users.noreply.github.com> * test(ollama): disable fixture key for keyless proxy guard --------- Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> Co-authored-by: Mrun25 <165297479+Mrun25@users.noreply.github.com> |
||
|
tavren27
|
826859588c
|
test(router): audit #41368 (#48794)
* test(router): cover silent_model shadow fan-out and streaming end to end Proxy cells check that each silent_model target gets one request in the caller's stream mode, that the caller sees only the primary, and that every shadow lands its own spend row under its own model group. SDK cells check the sync Router.completion thread path streams each shadow and fires its success callback. * test(router): flatten the spend row lookup with itertools.chain --------- Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> |
||
|
paveln83
|
7f44cff274
|
test(anthropic): audit #47478 (#48803)
Covers the json_schema envelope edge cases #47478 fixed on the Anthropic json tool path: a key listed only in `required`, a tool-named wrapper on a stream, non-ASCII and lone-surrogate answers, a stream cut before the tool block closes (proxy and sync SDK, Anthropic and Bedrock Invoke), and a string `output_format` with and without `drop_params` on Anthropic and Vertex chat and /v1/messages. The scripted Anthropic peer now refuses a non-object `output_format` the way Anthropic does Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> |
||
|
zoveki
|
53c6939ad4
|
test(anthropic, bedrock): audit #47489 (#48782)
* test(anthropic, bedrock): audit #47489 thinking replay across Claude routes and Bedrock Converse * test(bedrock): compare the whole replayed Converse turn in the merged-message and cache point cells --------- Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> |
||
|
norkelbyte
|
1176e608b1
|
test(harness): audit #47530 (#48751)
Check that cancelling a LocalSandbox run stops the grandchild process it started, using a real child process Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> |
||
|
tin-berri
|
8ced2cf433
|
fix(router): keep session tier pins coherent across workers (#48645) | ||
|
Kerry Lu
|
d6edd37810
|
fix(aiohttp_transport): close an owned aiohttp session when its transport is collected (#48650)
An OpenAI or Azure SDK client litellm built and evicted from the client cache is usually garbage collected inside the closer's 900 second grace window, so the closer's weak reference dies and the client's aiohttp session is never closed. LiteLLMAiohttpTransport.__del__ now hands an owned, still-open ClientSession to _close_recycled_session, which already covers the current-loop, foreign-loop and closed-loop cases. Co-authored-by: devin-ai-integration[bot] <158243242+devin-ai-integration[bot]@users.noreply.github.com> Co-authored-by: BangerSoul <bangersoul@users.noreply.github.com> |
||
|
devin-ai-integration[bot]
|
0657ed46cf
|
fix(langfuse_otel)!: derive trace-level name, input, output and tags on traces litellm starts (#42705)
* fix(langfuse_otel): derive trace-level name, input, output and tags on the generation span Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * fix(langfuse_otel): keep every reasoning summary in the derived observation output Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * fix(langfuse_otel): leave joined traces untouched by derived trace fields Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * fix(langfuse_otel): restore preset cache key fallback and type the derived-field helpers Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * fix(langfuse_otel): keep trace tags off joined traces entirely Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * test(langfuse_otel): tolerate out-of-band upstream requests and add sink outage chaos cell Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * test(integration): build the langfuse rig config without mutating the loaded yaml Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * refactor(langfuse_otel): flatten output items with itertools.chain to satisfy LIT014 budget Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * test(langfuse_otel): type the _RecordingSpan and _emitted unit helpers Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * fix(langfuse_otel): keep joined-trace tags, mask trace io, leave v2 trace io to the root Joined traces (existing_trace_id) keep the caller tags they carried before, and only the derived trace fields are skipped for them. mask_input and mask_output now redact the derived trace input and output, matching the vanilla Langfuse callback. The v2 mapper derives the trace name only, because the root observation already carries the trace input and output there. Uses the public preset cache key helper, moves imports to module top, restores the original formatting of the unit test file, and adds integration cells for the v2 llm_only scope and SDK sync/async calls. * fix(langfuse_otel): split string caller tags the way Langfuse reads them A caller tags value given as a comma list or a JSON array string (body metadata.tags or the langfuse_tags header) used to reach Langfuse raw, where ingestion split it into separate tags. Merging it with the derived tags kept it as one tag, so parse it the same way before the merge * test(otel): expect langfuse.trace.name on the v2 langfuse_otel generation span * fix(langfuse_otel): read string mask flags the way the classic callback does The langfuse_mask_input / langfuse_mask_output headers arrive as strings, so "false" redacted the derived trace io. Parse them with str_to_bool and fall back to truthiness, as the classic langfuse callback does * fix(langfuse_otel): derive trace fields only on traces litellm starts A litellm span that joins a trace an app or caller started (inbound traceparent, the app's active span, a caller-provided parent span) no longer gets a derived trace name, input, output or tags, like a request joining an existing_trace_id. Langfuse applies the later-starting span's trace update, so the derived values overwrote the app's own trace name and io. OTel v2 derives the name on the request root only when it has no parent, and the generation span keeps main's caller-only trace name. * test(langfuse_otel): keep main's OTel v2 generation key set in the arize family cell The OTel v2 generation span no longer carries a derived langfuse.trace.name, so the pinned key set goes back to main's. * fix(langfuse_otel): name the OTel v2 request root once, from the request's own call The v2 Langfuse logger re-derived langfuse.trace.name on the request root for every LLM call in the request. A post_call llm_as_a_judge guardrail's judge call replaced the caller's trace_name on the root with litellm-acompletion, and with another v2 callback (otel, arize_phoenix) listed before langfuse_otel that vendor's root span gained langfuse.trace.name on every request. The derived name now fills only an unnamed root this logger's provider started, and never comes from a call litellm makes for its own features. A caller trace_name still names the root on every call, as on main. * fix(langfuse_otel): keep the derived trace fields from pushing other attributes off a full span The derived trace name, tags, input and output were written after the OpenInference attributes with no room check, so a span near the SDK's attribute limit (about 34 tools at the default 128) evicted keys main exported, such as openinference.span.kind, metadata and llm.model_name, and raised the dropped count. They are now written last and only into the room the limit leaves next to the error attributes a failed call adds, so a full span exports what it would without them. * test(langfuse_otel): give the full-span cells' derived request a caller trace id, as the joined one's traceparent does * chore(langfuse_otel): drop the docstrings and comments the house rule disallows and type the new tests' parameters * test(langfuse_otel): set the span attribute limit the full-span tests run at and keep Final locals out of the app-trace loop --------- Co-authored-by: yucheng <yucheng@berri.ai> Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> Co-authored-by: yucheng-berri <293926326+yucheng-berri@users.noreply.github.com> |
||
|
yucheng-berri
|
715e9bb08d
|
fix(prometheus): count provider-side 401s in LLM failure and deployment metrics (#46668)
* fix(prometheus): count provider-side 401s in LLM failure and deployment metrics The invalid virtual key skip no longer applies to an error the LLM provider returned while the deployment was called with its own credential, so a 401 for a bad deployment credential increments litellm_llm_api_failed_requests_metric, litellm_deployment_failure_responses and litellm_deployment_total_requests and marks litellm_deployment_state as a partial outage again. Proxy-gate rejections and a failing credential that belongs to the caller (a rejected per-user credential, an api_key sent with the request, a per-user GitHub Copilot session) keep the skip * fix(prometheus): keep a caller's own Anthropic OAuth token failure out of the deployment metrics A provider 401 on an Anthropic OAuth token the caller forwarded through the proxy is the caller's credential failing, so it stays out of the deployment failure and outage series like a caller-supplied api_key. Adds integration cells for an Anthropic deployment's own key, the forwarded caller OAuth token and a direct SDK call that brings its own key * test(prometheus): pin how a caller's own key or Authorization header on a deployment is counted An SDK Router call that passes its own api_key is counted against the one-off deployment the router builds for that key, so the operator deployment keeps only its own failure. A provider 401 on an Authorization header the caller sends through the proxy counts against the deployment, as on v1.80.12 * fix(prometheus): keep caller-sent provider credentials and token-exchange 401s out of deployment metrics A caller can send its own provider credential in the request body, such as azure_ad_token, azure_scope, the Azure Entra app fields, watsonx token, gigachat_access_token, OCI keys, AWS keys or Vertex credentials, at the top level or inside extra_body. A provider 401 on that credential is the caller's failure, so it stays out of the shared deployment's failure, request, LLM-failed and state series, as on main. A token-exchange deployment trades the caller's identity token on every call, so its 401s stay out as well, matching the router cooldown. * test(prometheus): Final locals, precise helper types and one for per comprehension in the new test helpers * test(prometheus): pin that a provider 401 at a caller-sent api_base counts against its one-off deployment * fix(prometheus): keep a provider 401 on caller-sent headers out of deployment metrics A caller can override the deployment's credential with headers it sends in the request body: headers or provider_specific_header, or an Authorization, api-key, x-api-key, x-goog-api-key or Ocp-Apim-Subscription-Key entry in extra_headers. A provider 401 on such a request is the caller's failure, so it stays out of the shared deployment's failure, request, LLM-failed and state series, as on main. * fix(prometheus): skip a provider 401 only when the caller's headers carry a credential The proxy now records headers, extra_headers or provider_specific_header in credential_fields only when they name a provider credential header (Authorization, api-key, x-api-key, x-goog-api-key, Ocp-Apim-Subscription-Key), the same way api_key is recorded. A caller that sends a plain header such as a trace id still has a provider 401 on the deployment's own key counted against the deployment. * fix(prometheus): stamp credential_fields after the proxy forwards the caller's headers A provider credential header the proxy forwards from the caller's HTTP request (forward_llm_provider_auth_headers + forward_client_headers_to_llm_api) lands in data["headers"] after proxy_server_request is built, so the stamp missed it and a 401 on that caller credential was counted against the deployment. Compute the stamp after the header forwarding steps and pin both forwarded cases. * fix(prometheus): skip a provider 401 when the caller sends any header or an OpenAI organization The proxy now records headers, extra_headers or provider_specific_header in credential_fields whenever the caller sends or the proxy forwards any header there, and records organization when the caller sends one in the body or as a forwarded OpenAI-Organization header. A forwarded header can be a gateway key such as x-portkey-api-key or an OpenAI-Project id, and a wrong organization gets a 401 from OpenAI, so these 401s stay out of the deployment metrics as on main instead of marking a healthy deployment as a partial outage. * fix(prometheus): ignore the user information the proxy adds when stamping caller headers With add_user_information_to_llm_headers on and header forwarding enabled, the proxy adds x-litellm-user_api_key_* headers to the forwarded headers. Those carry the proxy's own key and user details, not the caller's, so they no longer make a request look caller-scoped, and a provider 401 on the deployment's own key on such a request is counted. * fix(prometheus): skip a 401 raised before the provider is called or at the caller's own endpoint A deployment with no key of its own fails with LiteLLM's local "Missing Anthropic API Key" 401 before any provider call, and a caller who sends aws_bedrock_runtime_endpoint (or another provider endpoint field) can get a 401 from their own gateway. Both were counted against the shared deployment and marked it as a partial outage, where main skipped every 401. Count a 401 only once the provider was called (api_call_start_time is set), and treat caller-sent provider endpoint fields like caller credential fields. * fix(prometheus): count the deployment's own 401 on requests that carry only anthropic-version or anthropic-beta The Anthropic SDK sends anthropic-version on every request, and the proxy copies anthropic-version and anthropic-beta into provider_specific_header for Anthropic, Bedrock and Vertex. Both were treated as caller headers, so a 401 on an Anthropic deployment's own key from any Anthropic SDK client stayed out of the deployment metrics. These two headers only pick the API version and beta features. Anthropic answers a bad anthropic-version or an unknown anthropic-beta with a 400, never a 401, so they cannot turn the deployment's own key into a 401 and no longer mark the request as carrying a caller header. * test(prometheus): read and write the owned proxy's own database in the expired-key and MCP cells Since #47213 an owned proxy boots on a private database, so the expired-key update and the MCP guardrail row read now go through the gateway's database_url instead of the worker's. * fix(prometheus): keep credential_fields as main stamps it and give Prometheus its own caller_credential_fields credential_fields tells the Rust OCR bridge which stripped body fields the caller sent. Repurposing it for the Prometheus caller-credential check changed its value (organization and extra_headers added, anthropic-version-only headers dropped). Restore main's stamp unchanged and stamp the Prometheus classification under caller_credential_fields, which only Prometheus reads. * fix(prometheus): skip a provider 401 on a stored credential the caller picks by litellm_credential_name A request body litellm_credential_name swaps the deployment's stored credential for one the caller chose, so a 401 on it is the caller's, not the deployment's. Treat it like the other caller credential fields so it no longer marks a healthy shared deployment as a partial outage. --------- Co-authored-by: yucheng-berri <293926326+yucheng-berri@users.noreply.github.com> |
||
|
yucheng-berri
|
7045139c3a
|
fix(router): stop concurrent deployment deletes from crashing deployment lookups (#47680)
* fix(router): stop concurrent deployment deletes from crashing deployment lookups delete_deployment and upsert_deployment shrink model_list before they repair the index maps, so a lookup on another thread in that window read a stale index: IndexError, or the neighbouring deployment and its credentials. An add racing a delete could corrupt the index for good. One RLock now covers model_list and the three index maps. Delete, upsert, add and the set_model_list reset change them under it, and every lookup that turns an index into a deployment reads under it. Fixes #45242 * test(router): release the held delete when the read reaches the index lock The race tests released the held write after a fixed 0.2 s, so a reader the runner stalled past it read the repaired list and passed on unfixed code. The read now runs on its own thread and the write stays held until the read returns or reaches the router's index lock; the 10 s bound only fails a broken synchronization. The integration test gains the audit cells: chat, streaming, Responses and Anthropic messages calls (sync and async, by model name and team public name) during a held delete, the same for upsert, add and set_model_list, a request for the deployment being deleted, a provider error, a double delete, and thread and async bursts during upsert churn. * test(router): cover the early return of a delete and the team filter of get_model_ids delete_deployment now returns None from inside the index lock, and get_model_ids by name reads its deployments through _deployments_named; pin both: deleting an unknown id returns None and leaves the list alone, and exclude_team_models drops only the team deployment. * fix(router): read deployments by name under the index lock outside the Router The response polling provider check, the complexity router's adaptive pool and the savings baseline turned the name index into list positions themselves, so a concurrent delete could still hand them a stale slot. They now read through Router.deployments_named, which takes the deployment index lock. * fix(router): type the locked deployment lookups as DeploymentTypedDict deployments_named and _team_deployments_named returned tuple[Any, ...]. They now return DeploymentTypedDict rows, which is what model_list holds. The adaptive pool setup in the Router and the complexity router drop their branches for pydantic Deployment rows, which these lookups never return, and should_include_deployment accepts the typed row. --------- Co-authored-by: yucheng-berri <293926326+yucheng-berri@users.noreply.github.com> |
||
|
Arsh Goyal
|
96792d0913
|
test(streaming): pin that a stream stuck on one chunk closes its provider stream and logs one failure (#38753)
* fix(streaming): stop re-raising and re-logging a failed stream on every next() A stream that raised keeps its provider iterator alive, so a consumer that iterates past the error re-detected, re-raised and re-logged the same failure on every call, flooding failure callbacks for one request. Latch the failure and re-raise it without logging again. * fix(streaming): drop the retained traceback before re-raising a failed stream * test(streaming): count failures through a real logging callback and require a clean stop on later reads * test(streaming): require the stuck stream to close early and release its socket The provider stream now records a GeneratorExit and caps its chunks, so draining it or detecting the loop late fails. Two wire cases check a sync and an async stuck stream release their TCP session. * test(streaming): avoid failure recorder name collision --------- Co-authored-by: Devin AI <devin-ai-integration[bot]@users.noreply.github.com> Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> Co-authored-by: moe-berri <moe@berri.ai> |
||
|
moe-berri
|
55479821e2
|
feat(router): add opt-in session repinning after successful failover (#48206)
* feat(router): add opt-in session repinning after successful failover * fix(router): respect shared session pins and split metadata * fix(router): preserve session routing during Redis failures * test(router): inject the delayed Redis reader |
||
|
Mateo Wang
|
c4360bbb59
|
fix(harness): set PWD to the working directory of local sandbox commands (#47530)
* fix(harness): set PWD to the working directory of local sandbox commands LocalSandbox started CLIs in the sandbox workdir but passed the parent's PWD through. OpenCode reads PWD as its project directory, so a litellm.agent() turn worked in the Python process cwd instead of the sandbox. * fix(harness): re-home resumed OpenCode sessions into the sandbox folder OpenCode files each session under the folder it ran in and only streams a resumed session's events from that folder, so a session saved before PWD pointed at the sandbox, or resumed in another sandbox, hung. Resume now runs opencode export and import in the sandbox first. Also types the CLI handler's config so its member accesses are known. * fix(harness): fail an OpenCode resume that cannot move its session, with a timeout The export and import step now has a 30 s bound, and a failure or timeout raises HarnessError instead of leaving the resumed turn waiting forever. A restart of a session the handler already ran skips the step. * fix(harness): resume an OpenCode session elsewhere as a copy and fail the turn when it cannot Importing the exported session moved it, so the original stopped working where it was saved. Resume now imports a copy under fresh ids, filed under the sandbox folder, and continues the copy. A failed or timed-out export or import ends the next turn with runtime_error, as base does for a missing session, instead of raising from agent_resume. * fix(harness): retry a failed OpenCode resume on the next turn and keep resumed copies stable Resume now only records the saved id, and the first turn runs the OpenCode export and import under the caller's timeout. A failed export fails that turn and the next turn tries again. The local sandbox kills a command when its run is cancelled, and the copy names the sandbox folder so OpenCode 1.15.5 and older keep it there. * test(harness): keep subprocess tests out of tests/unit The PWD regression test moves to tests/integration/sdk with an isolated child interpreter, and the cancel test checks collect_output with an in-process process instead of a real one. collect_output takes the sandbox Process protocol. * fix(harness): rewrite only id fields when copying an OpenCode session for resume |
||
|
Mateo Wang
|
91724bb0a7
|
fix(responses)!: count input tokens the way inference sends the request (#47363)
* fix(count-tokens): translate chat tool use before Claude token counts Responses and chat tool use (function tools, function_call items, tool messages) is now converted the way Claude chat inference converts it before the Anthropic, Azure AI and Vertex partner counters call count_tokens. Translation errors answer 400 before any provider call. Anthropic-shaped requests and /v1/messages/count_tokens keep their body as sent. * test(count-tokens): compare Claude count bodies with inference for Responses tool use * fix(responses): forward the caller's Responses body to OpenAI token counts OpenAI deployments now get the caller's /v1/responses/input_tokens body as sent, minus LiteLLM-only fields, so OpenAI's own count and errors come back. previous_response_id is checked for ownership like inference and routes to the deployment that served it * fix(responses)!: count Claude Responses bodies the way inference sends them Non-native Responses counts now run the body through the inference bridge, so text.format, tool_choice, reasoning, built-in tools, and file_search count the way inference sends them on Anthropic, Azure AI, Vertex, and Bedrock. The OpenAI native count body uses the drop_params-resolved tool fields. * fix(responses): honor drop_params and replay previous turns in Claude counts Claude input_tokens counts now read the same drop_params setting inference merges from the deployment, router defaults, and request, so a forced tool_choice downgrades to auto under drop_params. previous_response_id replays the stored turns through inference's session loader, and chat messages carrying fields beyond role and content go through the provider's chat translation * fix(responses): keep proxy-only request keys out of OpenAI counts and map chat tools in Bedrock counts * fix(count-tokens): import on Python 3.11 and keep stored prompts in OpenAI counts under drop_params * fix(responses): leave tool_choice out of counts when drop_params drops every tool * fix(count-tokens): count Mantle options, drop cache marks, and reject input inference cannot send The Bedrock Mantle fallback now counts thinking, tool_choice and output_config like the runtime count. Claude Responses counts leave out cache_control marks again, so more than four marks count instead of failing. Counts answer 400 before any provider call for user blocks chat inference refuses and for Responses tools with no Chat Completions equivalent, instead of dropping the block or counting a placeholder tool. Adds the wire tests for these paths * fix(count-tokens): count Responses tools the way the inference bridge sends them A Claude count now hoists additional_tools input items into tools, strips a malformed tool flag under drop_params and rejects it without, and leaves tools out when the deployment lists tools in additional_drop_params, so the count matches the body Claude inference sends for the same request * fix(count-tokens): keep Bedrock runtime counts option-free and send OpenAI the provider file id * test(count-tokens): capture the Anthropic count request at the HTTP layer instead of patching the handler * fix(count-tokens): ignore a caller-sent file id mapping and keep non-string file ids as sent |
||
|
Lai Wei
|
bda4e60f9b
|
fix(streaming): keep finish_reason and provider usage when rebuilding streamed text completions (#44349)
* fix(streaming): keep finish_reason and provider usage when rebuilding streamed text completions stream_chunk_builder_text_completion read finish_reason from chunks[-1] and recounted usage from messages. With include_usage the last chunk is a usage-only trailer, so finish_reason came back None (or raised IndexError when the trailer had no choices), and prompt_tokens was 0 because a text-completion prompt is not in messages. Scan for the last finish_reason and prefer the provider-reported usage, as the chat path already does. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(streaming): type the text completion chunk builder so its scans add no basedpyright debt Read chunk fields as typed attributes instead of untyped subscripts, which clears the reportUnknownArgumentType breach the lint gate flagged * fix(streaming): keep the provider's full usage, including token details, when rebuilding text completions Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(streaming): rebuild dict-form text completion chunks again stream_chunk_builder routes dict chunks holding TextChoices to the text completion builder, which now reads attributes. Convert them to TextCompletionResponse at the routing point. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix(streaming): accept dict chunks in the text rebuild and derive its usage like the chat path Moves the dict to TextCompletionResponse conversion into stream_chunk_builder_text_completion so direct callers keep working, and builds usage through ChunkProcessor.calculate_usage over the chunks that report counts, so a trailer without total_tokens keeps its prompt and completion counts and the result is a fresh Usage * fix(streaming): read text completion usage through the shared usage extractor so usage dicts on typed chunks rebuild * fix(streaming): merge text completion logprobs across chunks and rebuild dict text chunks as text completions * fix(streaming): rebuild one text completion choice per index and keep null logprobs fields null * fix(streaming): keep dumped chat chunks with a choice-level text on the chat rebuild path Route a dict chunk to the text completion builder only when its choice has `text` and no `delta`. The Responses bridge snapshots chat chunks with model_dump(), and Together's chat choices carry `text` next to `delta`, so streamed /v1/responses calls to Together ended without response.completed. Group text completion choices by index in one sorted pass, so rebuilding an n>1 stream costs about the same as rebuilding an n=1 stream of the same size. * fix(streaming): keep text completion rebuilds working on mixed chunks and unreadable logprobs A chat chunk inside a text completion stream now converts to a text chunk that keeps its finish_reason and adds no text. A dict chunk whose logprobs fail validation is read again with logprobs dropped, and logprobs that are not a Logprobs object keep the last value sent. Each of these inputs raised or lost data at the previous commit, and each now rebuilds without raising Also adds a Purview test showing a multi-choice text stream is blocked like the non-streamed response, and Final annotations on the new test locals * fix(streaming): rebuild text completion choices without an index at index 0 Text completion choices whose index is None now rebuild under index 0, as on the merge base, where the sort by index raised at the previous commit. A chat chunk inside a text completion stream now keeps the usage it reports, and a choice whose logprobs mix a Logprobs object with another shape keeps the last value sent, as on the merge base * fix(streaming): rebuild text completion choices with a string index under that int A text completion choice whose index is a string now rebuilds under the int it names, and one whose index cannot be read as an int rebuilds at index 0, as on the merge base. A string index mixed with an int index raised TypeError at the previous commit, and a string-only index came back as a string * test(streaming): drop a docstring that repeats its test * fix(streaming): fall back to choice 0 for an infinite text completion index A float infinity index made int() raise OverflowError, which the index coercion did not catch, so the rebuild failed where the merge base returned choice 0. --------- Co-authored-by: Lai Wei <8022184+roywei@users.noreply.github.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> |
||
|
Mateo Wang
|
2d181e6633
|
fix(token_counter): count DeepSeek V4 text with DeepSeek's own tokenizer (#47353)
Local token counts for DeepSeek V3 and V4 model names now use DeepSeek's own bundled tokenizer instead of cl100k_base, and tool definitions on V4 names count with V4's chat-template tool framing |
||
|
Vansh Singh
|
f4347d4f50
|
fix(speech)!: correctly propagate spend logging and usage for Gemini TTS bridge (#44551)
* fix(speech): correctly propagate spend logging and usage for Gemini TTS bridge * fix(speech): call Gemini once per TTS request and log the bridge's token usage aspeech re-ran speech whenever the first result was not a coroutine, so the speech-to-completion bridge sent every Gemini TTS request upstream twice. It now returns the sync result. The bridge's inner completion runs as an internal sub call, so only the speech call fires success callbacks, and the completion's usage rides on the audio response so the spend log and usage_object carry real token counts. Replaces the separate inner Logging object and the cost_calculator fallback, which never reached the spend row. * fix(speech): flag a cached Gemini TTS reply as a cache hit so it logs at no charge * fix(speech): price a cached Gemini TTS reply's savings from its token usage and keep its cache key * test(integration): cover Gemini TTS speech billing, logging, cache hits and deployment TPM * test(integration): start the speech TPM cell in a minute's first 10 s so the two requests never cross a minute --------- Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> |
||
|
yucheng-berri
|
ca0ec627b6
|
fix(exa): return highlights, highlight scores, score and the response envelope from Exa search (#47423)
Co-authored-by: yucheng-berri <293926326+yucheng-berri@users.noreply.github.com> |
||
|
Mateo Wang
|
52f5a28679
|
fix(bedrock): send native structured output first on Opus 4.8 and Sonnet 5 and fall back to the json tool when Bedrock rejects it (#47100)
* fix(bedrock): send native structured output first on Opus 4.8 and Sonnet 5 and fall back to the json tool when Bedrock rejects it
* fix(bedrock): retry with the json tool when Bedrock refuses the native output schema
Count a 400 under output_config.format.schema: as a native output rejection, so a schema Bedrock cannot compile (like a pattern with a nested {n,m} group) gets the json tool retry instead of reaching the caller
* fix(bedrock): make the native output retry send what an opted-out deployment sends
The retry after a native output rejection now reruns the same optional param
mapping completion() ran, with the deployment model_info opted out of
supports_bedrock_native_structured_output, so tool_choice, effort and the json
tool schema match an opted-out deployment instead of a hand-built tool. Each
config keeps only the predicate that decides when native output came from the
opt-in key.
Bedrock batch records for Anthropic models map with the opt-in turned off,
since a batch job has no retry. Models on supports_native_structured_output
keep native output in batch
* fix(bedrock): build the messages retry from the caller's request and keep other errors unchanged
* fix(bedrock): give sync converse stream errors the upstream body as text
* fix(bedrock): let a malformed caller outputConfig reach Bedrock instead of raising
A Converse json_schema request whose caller outputConfig failed litellm's
shape check raised a pydantic error before the request was sent, so the
caller got a 500 where the merge base passed Bedrock's 400 through. The
opt-in check now treats that outputConfig as carrying no native output.
Also opts the vertex wire test's pinned Opus 4.8 json tool deployment out of
native output, and extends the Bedrock native output integration tests.
* fix(bedrock): send a refused caller outputConfig once and give opt-out advice only when native output is unsupported
A caller can send its own outputConfig on Converse. When Bedrock refused that
schema on an opt-in deployment, the retry still carried the caller's textFormat,
so Bedrock got the same request twice and the warning told the operator to turn
off an opt-in that was not the cause. The retry is now skipped when the opted-out
request would still carry a native textFormat, so Bedrock's 400 reaches the
caller after one call as on the merge base. The opt-out advice is logged only for
the account-level "Extra inputs are not permitted" rejection.
* fix(bedrock): keep the caller's continuation message on the native output retry
The Converse transform pops user_continue_message from litellm_params on the first attempt, so a retry built from those params afterwards dropped it. Both Bedrock chat routes now hand the fallback a copy taken before the first attempt.
* test(bedrock): pin additional_drop_params on the /v1/messages native output retry
* fix(bedrock): give the Converse native output retry the fake stream timeout
|
||
|
Mateo Wang
|
c143954493
|
fix(streaming): report the deployment's api_base when a stream fails mid-way (#47045)
* fix(streaming): report the deployment's api_base when a stream fails mid-way * fix(responses): report the deployment's api_base when a native responses stream fails mid-way * fix(streaming): name the deployment's configured api_base in mid-stream errors and keep prompt text out Mid-stream errors on chat and native /v1/responses streams now prefer the router's configured api_base over the logged request URL, websocket turns keep the connection's api_base, and the mid-stream mapping no longer adds the prompt to the debug info. * fix(streaming): keep api_base credentials out of error debug info and leave websocket turn logging unchanged * fix(streaming): name the websocket connection's api_base on turn failures and strip scheme-less userinfo * fix(streaming): strip api_base userinfo before cutting the query so a raw ? or # in a password stays out Adds the audit's integration tests for the mid-stream alert API Base line and the SDK debug info. * fix(streaming): leave the API Base line out when an @ follows the host A raw / ? or # in a password and an @ in a query or path read the same in a malformed api_base, so either guess could show a secret or a wrong host. * fix(streaming): keep the API Base line when an @ sits in a provider path Vertex partner model ids like claude-opus-5-5@default put an @ in the URL path, so the fail-closed rule left out the line main printed. The line is now left out only when the text before the first / ? or # is not a host, and a :// counts as the scheme only after a valid scheme name. * test(streaming): cover a numeric password with a slash in the API Base line The audit found that a password like 1234/function-secret, which looks like a port, had no case. The line is left out on this branch and printed the raw secret on main |
||
|
Mateo Wang
|
2c5c9b8d84
|
fix(bedrock): use native structured output on Haiku 5.5 and Sonnet 5.5, and fix Fable 5 and Sonnet 4.6 schema 400s (#46870)
* fix(bedrock): use native structured output on every Claude model whose Bedrock profiles accept it Bedrock accepts outputConfig structured output on every reachable profile of Claude Opus 4.7, 4.8, 5, 5.5, Sonnet 5, 5.5, Haiku 5.5, and Fable 5.1, so json_schema now goes out natively there instead of as a forced tool, which kept Bedrock from returning adaptive thinking. Fable 5 rejects outputConfig, so its keys go back to the tool path. GovCloud keys keep their flags, and the drift tests let GovCloud lag its commercial base but never lead it. * fix(bedrock): clean Claude schemas for native structured output and keep recursive schemas on the old path Bedrock native output rejects minimum/maximum, minItems, uniqueItems, oneOf, additionalProperties true, and recursive or unresolved $ref, all of which the forced tool accepted. Claude schemas on the Bedrock native paths now go through the cleaner the Anthropic native path already uses, and a schema with a recursive or unresolved $ref keeps the forced tool, inline, or drop path it had. * fix(bedrock): keep older installs on the json tool with a Bedrock-only native output key Older installs fetch the model map from main, so flipping supports_native_structured_output moved them onto native output with their old schema handling. The new supports_bedrock_native_structured_output key is ignored by older code. Schemas native output cannot hold (recursive refs, open objects such as dict fields) keep the json tool on the new key, models on the old key keep their current behavior for those shapes, and the Anthropic API, Vertex AI, and Azure keep their current behavior. * chore(cost-map): regenerate the model prices schema for the Bedrock native output key * fix(bedrock): keep native output to profiles every account accepts and route schemas Bedrock rejects to the json tool Only Opus 5.5, Sonnet 5.5, Haiku 5.5 and Fable 5.1 keep the Bedrock native output key, since Opus 4.8 and Sonnet 5 answer 400 on some accounts and Opus 5 garbles escaped characters. Lookaround patterns and string formats outside Bedrock's list now keep the json tool, objects typed as a union with null are closed, and the cleaner keeps minItems of 0 or 1. * fix(bedrock): keep Opus 5.5 and Fable 5.1 on the json tool, since native output stops their adaptive thinking * fix(bedrock): keep shared $defs in native schemas and send refused schemas to the json tool on every native model Native output now keeps $defs instead of inlining each use, so a submodel reused across a schema counts once against Bedrock's optional and union limits. Schemas native output cannot hold use the json tool on Converse and Invoke chat for both native keys, and minItems is only kept when it is an integer 0 or 1 * fix(bedrock): keep the json tool unless thinking is requested and bound native output schemas The Bedrock native output key now applies only when the call asks for thinking, so calls without thinking keep the json tool as before. Schemas past Bedrock's documented union and optional counts, or past a grammar size estimate fitted to live probes, keep the json tool on both map keys. Open objects keep the json tool on the Anthropic API, Vertex AI and Azure AI, where native output returned them empty. The $ref check no longer expands $defs, so doubling definitions no longer stall the request. * fix(bedrock): resolve root-path $defs refs so nested Pydantic models keep native output on Invoke * test(bedrock): cover native structured output and json tool routing through the proxy Integration cells drive the proxy against a scripted Bedrock peer that enforces Bedrock's structured output schema rules, covering thinking with native output on Converse, Converse streaming, Invoke, /v1/responses and /v1/messages, the deployment opt-in, Fable 5, schema cleaning on every native entry, open objects, and malformed schemas. Vertex, Anthropic and Azure AI cells cover open objects on native-capable models and the cleaner's minItems and nullable object handling. The unit table now covers reasoning_effort none and disabled thinking * fix(bedrock): keep native output for numeric formats and wide old-key schemas, and keep thinking in json mode Check string formats only on nodes that can be strings, apply the 64-field cap only to the Bedrock-only key, carry thinking blocks into the json mode message, and send malformed json tool schemas as object schemas. * fix(anthropic): keep native output for dict fields unless the json tool can be forced * fix(anthropic): keep native output for dict fields when the caller sends tools * test(bedrock): close the dangerous-tool-use test schema, since Bedrock native output requires closed objects * test(anthropic): cover forced and unforced open-object routing, json mode thinking, numeric formats and wide schemas Add proxy cells for open-object schemas that keep native output when the json tool cannot be forced (a model that refuses forced tools on the Anthropic API and Vertex AI, thinking, caller tools), for the thinking block a json mode answer keeps on Vertex AI and Bedrock Invoke, and for integer and number formats and 65-field schemas on each Bedrock map key. Point the forced open-object cells at the Anthropic API and Azure AI only, since no Vertex AI model in the cost map takes both native output and a forced tool * fix(bedrock): keep the native output opt-in on the deployment that sets it, and route effort none open objects to native output * test(bedrock): cover deployment-scoped native output opt-in, register_model and effort none open objects |
||
|
Kerry Lu
|
1451aed726
|
fix(tests): drain SDK logging before restoring callbacks (#46924)
Co-authored-by: devin-ai-integration[bot] <158243242+devin-ai-integration[bot]@users.noreply.github.com> |
||
|
Mateo Wang
|
7ddc9f1c96
|
fix(bedrock)!: keep reasoning, thinking blocks and the json_mode finish_reason on fake streams (#46863)
* fix(bedrock): keep reasoning, thinking blocks and the json_mode finish_reason on fake streams A Bedrock Converse model without native structured output fakes the stream when response_format carries a json_schema. The fake stream transformed the response without json_mode and rebuilt a chunk from text, tool_use and token counts only, so it dropped reasoning_content, the signed thinking_blocks and the reasoning token count, and ended on finish_reason tool_calls. The three fake-stream sites now pass json_mode to _transform_response and stream the resulting ModelResponse through the shared convert_model_response_to_streaming converter, which now numbers each streamed thinking block with its index the way real streams do * fix(bedrock): parse text tool calls on Converse fake streams like the non-stream answer * fix(responses): stream the answer text when one chat chunk carries reasoning and the answer * fix(streaming): keep every field on the OpenAI and openai_like fake streams and split multi-block thinking Azure o-series and openai_like fake streams now use the shared MockResponseIterator, so extra tool calls, reasoning and cached and reasoning tokens survive; the Bedrock-local iterator is deleted. The /v1/responses bridge also splits a chunk whose thinking block is signed or redacted with no reasoning text, and the sync iterator closes the reasoning item before the answer opens, like the async one. The /v1/messages bridge gives each thinking block in a chunk its own content block, starts a new block when the thinking block index changes, and streams redacted thinking as redacted_thinking. Snowflake replays drop the stream index from thinking blocks. * fix(anthropic): stream each thinking block's text once on /v1/messages An unsigned thinking chunk now opens its content block with empty text, since its thinking_delta carries the text. A multi-block chunk keeps each piece's thinking block, so an unsigned block followed by a signed one keeps its own text and index. * docs(anthropic): describe which thinking block start carries its text * fix(streaming): close merged reasoning before the answer when one chunk carries both * test(streaming): cover fake-stream reasoning across the proxy bridges and type the Snowflake replay test |
||
|
ryan-crabbe-berri
|
977a581213
|
fix(router): stop queueing budget increments for Redis when no Redis is configured (#46831)
* fix(router): stop queueing budget increments for Redis when no Redis is configured Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> * test(proxy): assert Redis-less model budget limiter tracks spend locally without queueing Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> --------- Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> |
||
|
yuneng-jiang
|
b602209ca6
|
test(integration): hand every harness server a port from one per-worker block (#46761)
* test(integration): hand every harness server a port from one per-worker block
About 45 places in the integration harness picked a port by binding port 0,
reading the number back, closing the socket and starting a server on that
number later. Another process could take the port in between, which surfaced
as "Address already in use" flakes. Earlier fixes retried around the race in
single call sites without removing it.
tests/integration/_support/ports.py now owns every port the harness picks
ahead of a server. Each pytest process claims a 200-port block between 10000
and 32767 by taking an exclusive flock on a per-block lock file in the system
temp directory, so parallel runs on one machine never share a block and the
lock goes away with the process. The range is above the fixed service ports
and below the Linux and macOS ephemeral ranges. Ports go out in order, are
never reused in a run, skip anything an outside process already holds, and
a process that exhausts its block claims the next free one.
The three _free_port copies, worker_endpoints' port-0 binds,
free_loopback_port, the owned proxy's lost-port-race retry and the owned
Redis retry loop are gone. Tests that restart a server on the same port now
allocate that port up front instead of reading it back from the first
server. TQ012 in scripts/check_test_quality.py rejects a port-0 bind read
back with getsockname anywhere under tests/integration.
* ci(integration): record listening sockets and the ephemeral port range
* Revert "ci(integration): record listening sockets and the ephemeral port range"
This reverts commit
|
||
|
Mateo Wang
|
7090588bdf
|
feat(gemini)!: auto-continue gemini-4-argon long decodes with continuationToken (#46240)
* feat(gemini): continue MAX_TOKENS decodes with continuationToken across chat, streaming, and interactions * fix(gemini): continue long decode on CONTINUATION and cancel interactions at /cancel * feat(gemini): add public gemini-4-argon cost-map rows for AI Studio and Vertex AI Add gemini/gemini-4-argon and vertex_ai/gemini-4-argon alongside the existing early-access row, carrying the same limits and capabilities. The AI Studio row mirrors the early-access entry; the Vertex row differs only in provider and pricing source. Prices follow once Google publishes them. Extend the mapped metadata test to cover all three ids, that each routes to its provider, that its backup copy matches, and that the public rows carry the same limits as the early-access row. * feat(gemini): raise gemini-4-argon to the 1M long decode cap and report initial_prompt_tokens across hops * chore(prices): build on the merged gemini-4-argon rows and only raise the long decode cap * fix(gemini): follow continuation hops in a loop instead of recursion The unary hop collectors and the stream iterator called themselves once per continuation request, which the recursion check in CI rejects and which would hit the interpreter's call depth on a very long output. They now loop, and a test follows more hops than the interpreter allows nested calls * fix(gemini): bill Google Search grounding from every continuation request Google sends search grounding once, on the request that ran the search, and does not repeat it on a continuation request. The merged response kept only the last request's grounding, so usage.web_search_requests and the per-search fee dropped every earlier search. Unary merges and the stream iterator now carry the grounding of every request in the chain * fix(gemini): continue long gemini-4-argon answers on streams and Vertex AI A continued stream stopped after the first chunk of its second request. The stream wrapper asks the iterator for a fresh start before every chunk, and that reset it to the first request's finished stream. The iterator now reads from the newest request's stream On Vertex AI the continuation request went to the v1 API version, which rejects continuationToken with a 400. A request that carries the token now goes to v1beta1, which accepts it, including a token that v1 issued On a stream, initial_prompt_tokens came from a usage chunk in the middle of the stream, which reports a smaller prompt count than the last chunk. It is now the count the first request's stream ends with, the same number a unary call reports continuation_token is no longer forwarded on the Interactions API, which answers 400 for the field. It is dropped there as it was before this branch * chore(gemini): name the type errors the dual-mode stream iterator trips The stream iterator holds a sync or an async line iterator in one untyped attribute. Pointing it at a continuation request's stream gave the type checker a concrete union for it, so six existing lines that only ever run in one of the two modes started to report errors This adds a rule-named suppression with a reason on each of those lines and changes no code. Splitting the class into a sync and an async iterator would remove them and is left for a follow-up * fix(gemini): continue only when Google ends a request with CONTINUATION, and stop streaming mid-answer tokens A long decode now sends the next request only when Google's finish reason is CONTINUATION, so an answer cut at the caller's own max_tokens returns in one request with its continuation_token instead of drawing a rejected extra request A request for more than one choice is never continued, since Google rejects a continuation with more than one candidate Chat-format streams no longer carry the tokens Google sends mid-answer. Only a stream that ends cut off carries a continuation_token, on its last content chunk The response cache key now includes continuation_token, so a resend with a token is not served the earlier segment * fix(gemini): keep one response id across a continued answer A continued chat stream changed its chunk id each time the next request to Google started, and a unary answer came back under the last request's id. Every chunk and the unary answer now carry the first request's id, including the chunk that ends a cut-off stream A request whose n or max_tokens arrives as a string is sent once and never continued, where it used to fail with a 500 on the comparison. An empty continuation_token no longer changes the cache key * test(gemini): add integration cells for long decode continuation Covers continued answers on chat, /v1/messages and /v1/responses, streamed and not, on the Gemini API and Vertex AI against a scripted Google, plus concurrency, outage and worker restart runs Also moves the Google stand-in cancel route to /cancel, the URL this PR now calls for Interactions cancel, and checks the cancel test posts there exactly once * fix(vertex_ai): fetch the access token again before each continuation request A long Gemini 4 answer on Vertex AI can run for over half an hour of follow-up requests, and every one of them reused the bearer token fetched before the first request. Once that token expired Google answered 401 and the caller got a cut-off answer with finish_reason length Each Google request in a chain now asks for the token again, which is a cache hit until the token nears expiry, on the streamed and unary paths, sync and async. A refresh that fails counts as a failed follow-up, so the caller still gets the answer so far with its continuation token. A caller's own Authorization in extra_headers still wins on every request, and the Gemini API key path is unchanged * fix(vertex_ai): start each continuation stream with an empty parser buffer When a Google stream ended on an unfinished frame, its leftover text stayed in the stream parser and every frame of the next continuation stream was appended to it. None of them could be decoded, so the whole continuation was billed and dropped without an error Each continuation stream now starts with a fresh buffer, and a test covers a stream that ends on an unfinished frame, sync and async * fix(vertex_ai): treat a continuation stream with no readable frame as a failed follow-up A continuation stream that answered 200 and then sent Google's error frame first, sent a body that is not JSON, or dropped before its first frame ended the caller's stream with `stop` or an error and no token, so a cut answer looked complete or was lost. It now ends the way a refused follow-up does, with the answer so far, `finish_reason` length and that request's token Every continuation stream request now also gets the caller's per-request timeout, which #46349 added to the first one, so a follow-up that never answers ends at that timeout the same way * fix(vertex_ai): keep every follow-up's citations and URL context in a continued answer A non-streamed answer that took several Google requests kept citationMetadata and urlContextMetadata from the last request only, and left every later request's offsets counting from that request's own text. So a URL read on the first request was missing from the answer, citations from earlier requests were dropped, and a later request's citations and search supports pointed at the wrong part of the joined answer The joined answer now keeps every request's citations (Vertex AI's `citations` and AI Studio's `citationSources`) and URL context, and shifts each later request's citation and search support offsets by the UTF-8 bytes of the answer text sent before it, thought parts excluded, since Google counts these offsets in bytes * fix(vertex_ai): answer when a candidate sends null citationMetadata or urlContextMetadata A Gemini candidate whose citationMetadata or urlContextMetadata was JSON null made the call fail on every Gemini model, non-streamed with a 422 or 500 and streamed with an error event, because the per-request metadata helper handed the null back as is and the caller tried to extend a list with it. A continued answer also failed when a follow-up's citation or search support came with a null startIndex A null field is now kept as one entry on a single request, as it was before this branch, a follow-up's null URL context adds nothing to a continued answer, and a null startIndex counts from the start of that request's text. Tests cover missing, null and empty fields on one request and on a continued answer, streamed and not * fix(vertex_ai): pass through spans with no end index and point a continued stream's citations into the whole answer A later request's citation or search support with no endIndex made the whole non-streamed answer fail with a 500, so a span Google sends without an end index now comes back as Google sent it, the way one request for the whole answer returns it. A span with an end index but no start index still counts from 0 A continued stream handed callers each later request's citation and search support offsets counted from that request's own text. The stream now adds the answer bytes of every earlier request, so the offsets point into the whole answer like the non-streamed path and one request for the whole answer Adds the audit's wire tests for both, which fail 24 cells before this change * test(vertex_ai): cover a later request whose cited span starts at its first byte with no start index Google leaves out a zero startIndex, so the second request of a continued answer can cite its own opening with only an endIndex. These wire cells check that span lands where one request for the whole answer puts it, on Gemini and Vertex AI, chat and Responses, unary and streamed * fix(vertex_ai): log a continued answer's raw response as one object A non-streamed answer joined from several Google requests logged its raw response as a JSON list of every request's body. The OpenTelemetry logger reads that raw response as one Gemini body to set its `llm.gemini.*` attributes, so on a continued answer it hit a validation error and set none of them It now logs the merged body, which carries only the last request's token, and a single request still logs Google's body unchanged. The new unit test fails with the change reverted Also adds integration cells for a cut chat stream's token chunk, an answer joined from 41 requests, a caller that gave up on a non-streamed call, counts and caps that are not numbers or are sent twice, `n` sent as true, a partial answer cached after a refused continuation, the token in another provider's cache key, and the sync SDK path with its Vertex token refetch * test(integration): check a continued answer logs one merged raw response A non-streamed Gemini answer that took several Google requests now hands logging one merged response, so the OpenTelemetry span carries the llm.gemini attributes. These 12 cells fail on the merge base and on the commit before that fix, and pass on the branch * test(vertex_ai): cover a follow-up tool call and its thought signature in long decode * test(gemini): pin that a one-request answer logs Google's body unchanged |
||
|
Mateo Wang
|
62af39aaf8
|
fix(anthropic): count a replayed server tool call's cache mark the way modify_params sends it (#46591)
* fix(anthropic): count a replayed server tool call's cache mark the way modify_params sends it With modify_params on, a trailing tool search turn whose call has a kept result got a placeholder answer from the factory and went out as a marked tool_use, but the cache control hook counted marks before that placeholder existed and skipped the mark, so a request at the cap reached Anthropic with 5 marks and failed with a 400. The hook now finds client answers on the sanitized messages, the same list the factory reads. Also make the replay checks cheaper (prefix check before validation, early return without provider_specific_fields, and look up client answers only when a tool call carries cache_control), drop the two audit cells that cover old gaps outside the tool search fix, and derive the pass-through test timeout from the graceful stop budget. * fix(anthropic): keep the duplicate tool result warning out of the cache control hook The hook finds client answers with only the step of the modify_params sanitize pass that pairs tool calls with their results, so a request with a duplicate tool result logs the warning once again instead of once per mark count. Also restore the Snowflake non-stream and cached-stream tool search cells as known bugs with pytest.skip instead of deleting them, and put the Anthropic pass-through test timeout back to 240 seconds. * fix(anthropic): count cache marks next to a message the pairing step cannot read With modify_params on and a marked tool call, the cache control hook paired tool calls with their results before counting. A message without a get method (a stray string or null in extra_body.messages, or an openai SDK ChatCompletionMessage) made that pairing raise AttributeError before any provider call, where main counted the marks and let the request through. The hook now pairs only when every message can be read the way the pairing step reads it, and otherwise counts on the messages as sent, like main did. * fix(anthropic): count cache marks after placing a later system message the way the request builder does * test(anthropic): cover a leading system prompt and a system message before the answer in the cache mark census * test(anthropic): align the tool search replay cells with main and cover a system message after the search turn The malformed kept result cells now expect the unanswered search call to be dropped, as main does since #45882. The stream cut cell checks streamed tool call names, since the kept search result names get_weather as a tool reference. The streamed response cache hit cell runs again because #46104 fixed it. The modify_params cap cell expects the four client marks, since the placeholder answer sends the search call as a plain tool_use with its mark, and a new cell covers a system message after the trailing search turn in both stream modes * fix(anthropic): count marks on the raw messages when they cannot be placed, and find client answers once per request * test(anthropic): cover marked tool calls next to a malformed tool id and a later system message * perf(anthropic): look up client tool answers only when a marked tool call is a server tool call Answered ids only decide whether a marked srvtoolu_ call is rebuilt as server_tool_use, so a request whose marked tool calls are all client calls no longer places system messages or runs the modify_params pairing step to count marks * test(anthropic): reach the client answer lookup with a marked search call in the malformed tool id cell * perf(anthropic): place mid-conversation system messages in linear time for models that keep them * fix(anthropic): count cache marks with the model's own system message placement and pair tool results in linear time * test(anthropic): cover modify_params dropping a tool answer that answers no call of the latest assistant turn * fix(anthropic): count marks with in-place system placement when the model lookup cannot confirm the flag * test(anthropic): cover a bare model with custom_llm_provider filling the cache mark budget through litellm.completion and Router * fix(anthropic): give unmapped models the request builder's converting placement and survive any model lookup failure * test(anthropic): cover an unmapped model name counting cache marks with the request builder's converting placement * fix(anthropic): count server tool call cache marks the request builder's way only on routes that build Anthropic Messages requests Databricks, OpenRouter and Bedrock Converse send every marked tool call as is, so the hook keeps the census main uses there. The seeding step carries custom_llm_provider on the injection points when a server tool call is marked, so a bare model name still resolves its route. * test(anthropic): cover two answered marked server tool calls before a system message on Databricks, OpenRouter and Bedrock Converse * fix(anthropic): keep main's cache mark count on GitHub Copilot and ChatGPT routes without resolving them, so a sync call with no stored login never starts the device flow * test(anthropic): cover a marked server tool call reaching the GitHub Copilot and ChatGPT login hosts like an unmarked one |
||
|
Kerry Lu
|
7faa1a24bf
|
test: extend runtime regression contracts for Messages and SDK (#46641)
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> |
||
|
Taylor Hawkes
|
fd84e97f49
|
feat(providers): add Inference APIs as OpenAI-compatible provider (#41669)
* feat(providers): add Inference APIs as OpenAI-compatible provider Registers https://api.inferenceapis.com/v1 as a JSON-configured provider (chat completions and Responses API). Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> * fix(providers): list Inference APIs in Add Model and the provider docs index, add request-level tests - provider_create_fields.json entry so the Add Model dropdown test passes - provider_endpoints_support.json (+ backup) entry so the provider docs check passes - tests now capture the outgoing HTTP request (URL, auth header, body) instead of only inspecting registrations - drop a redundant comment Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> * test(providers): inject a recording HTTP session instead of patching httpx.Client Request-level tests now set litellm.client_session to an httpx.Client with a MockTransport, so nothing is monkeypatched on a class. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011qUBY2P8qTTuJCUeMzTCD4 * chore(providers): drop redundant comment on the inferenceapis entry Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_011qUBY2P8qTTuJCUeMzTCD4 * feat(inferenceapis): add cost map rows for the 12 chat models and test the responses route Prices and context windows come from https://inferenceapis.com/pricing and GET /v1/models. Moves the provider test to tests/unit and adds the interactions flag the endpoint matrix now carries. * feat(inferenceapis): price the vendor's chat model aliases and mark /v1/messages supported * fix(inferenceapis): serve /v1/messages natively and keep audio routes off the provider * fix(json providers): send provider-specific params in extra_body and report context overflow as ContextWindowExceededError * fix(exception mapping): match the Inference APIs overflow phrase instead of any context_length_exceeded substring * fix(exception mapping): map context overflow to ContextWindowExceededError for Inference APIs only Providers without their own mapper keep reporting an overflow as BadRequestError, since the router raises a ContextWindowExceededError instead of trying generic fallbacks when context_window_fallbacks has no entry for the group. Adds the integration cells for the provider, its fallbacks, and the extra_body and overflow families. * test(inferenceapis): hold upstream replies across the worker kill in the chaos cell A reply the killed worker sent before SIGKILL can lose its queued spend row, which the product never promised to keep. Holding the upstream until after the kill means only the surviving worker answers, so every 200 must log exactly once. * test(inferenceapis): make the chaos cell kill the quieter worker and always release held replies The survivor now must hold at least one in-flight request, so the burst check cannot pass as 0 == 0, and the held vendor is released on exit so a failed assertion cannot stall the wire server teardown --------- Co-authored-by: TaylorHawkes <857858+TaylorHawkes@users.noreply.github.com> Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com> Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> |
||
|
구동하
|
a9e45a24d3
|
fix(streaming): keep provider response headers on SDK streaming errors (#41528)
* fix(streaming): keep provider headers on MidStreamFallbackError A 429 raised while opening a stream is wrapped in MidStreamFallbackError, which dropped the litellm_response_headers exception_type() attaches to the wrapped error. retry-after and x-request-id reached callers on the non-streaming path but not on the streaming one, so retry backoff, provider request id logging and proxy error headers lost them. * refactor(streaming): drop explanatory comment from MidStreamFallbackError header copy * test(streaming): cover sync and async header carry on MidStreamFallbackError Parametrizes the provider-header test over the sync and async stream wrappers, and drops the hasattr test, which pinned how the attribute is stored rather than what any caller observes * test(streaming): add integration cells for stream-open provider headers Cover the SDK stream openers (gemini, vertex_ai, codestral, predibase and the eager providers), the Router stream retry and fallback paths, and the proxy on both the router and the config-less serving paths, including a mixed burst, a worker SIGKILL and a SIGTERM restart. Add a helper that boots a config-less proxy owned by the test --------- Co-authored-by: mateo-berri <277851410+mateo-berri@users.noreply.github.com> |