The oniguruma run of the ByteLevel pre-tokenizer regex is about 90% of
encode_fast on a 100k token body (100 ms of the ~110 ms Rust admission
count in the gateway pod). A hand-written scanner that yields the same
pieces, then feeds the model directly, counts the same text in 10 ms.
It only engages for tokenizers with the Anthropic shape (optional NFKC,
ByteLevel without prefix space, no post-processor) and falls back to the
full encoder when the text contains an added token. Parity with
encode_fast is tested on random texts, the pieces are compared with the
real pre-tokenizer, and the \p{L}/\p{N}/\s tables are checked against
oniguruma for every code point.
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Profiling the sidecar-enabled gateway at 700 rps showed ~2.4% of all samples
in get_model_group_info called per request from budget reservation, plus
get_deployment_model_info for tiered pricing tables. Both are read-only lookups
over the model list, so serve them from the Router's lru caches and clear the
deployment cache alongside the group cache when the model list changes.
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Inference workers publish one compact typed SpendEvent per success over a
unix socket or loopback TCP; an opt-in sidecar (python -m gateway.spend_worker)
runs the unchanged _ProxyDBLogger cost pipeline against the pod's pgbouncer.
Default off (LITELLM_SPEND_WORKER_ENABLED). Also reuses the preset cache key
in get_logging_payload instead of hashing the request body
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Adds a pgbouncer build stage to gateway/Dockerfile and a gateway.launch
supervisor that assembles DATABASE_URL, starts PgBouncer once per pod when
LITELLM_PGBOUNCER_ENABLED is set, and then runs uvicorn with the loopback URL
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
get_candidate_model_ids_for_route (added in #40280 for the encrypted-content affinity
check) reconstructed the candidate pool by unioning the model_name and team indexes with
pattern_router.route. That diverged from how the router actually resolves a route: it took
a union instead of the first matching path, and pattern_router.route only matches the
literal name, so a provider-qualified pattern (matched by get_deployments_by_pattern, which
retries the {provider}/{model} form) was missed and the default deployment was ignored.
For an affinity follow-up on a wildcard or team-public route, that mismatch could strip
encrypted reasoning on a same-group cooldown, or return a 503 on a real cross-path switch.
Delegate the non-model_name case to _try_early_resolve_deployments_for_model_not_in_names,
the same resolver _common_checks_available_deployment uses, so candidate membership follows
the router's real precedence. With include_team_models left off it stays read-only and does
not raise. Behavior for concrete model groups and routing groups is unchanged.
Claude-Session: https://claude.ai/code/session_01KAumQbhzk6jdWWHFLA8Jar
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
The FK cascade drops the LiteLLM_JWTKeyMapping row, but the cached
jwt_key_mapping:{claim}:{value} entry still resolved to the deleted token
hash, so every JWT call from that identity failed until
virtual_key_mapping_cache_ttl expired instead of auto-registering against a
recreated key. delete_verification_tokens now snapshots the mapping cache
keys before the delete and evicts them across replicas afterwards, the same
way /key/regenerate already does.
Claude-Session: https://claude.ai/code/session_011Tn3657NkV6ojLqewL64Kb
* fix(langfuse): give session-header calls their own trace id
A client that sends only a session header (x-litellm-session-id, a vendor
x-<name>-session-id such as Claude Code's X-Claude-Code-Session-Id, a bare
x-session-id, or the Codex session/thread/conversation family) has that value
stamped into both trace_id and session_id by the proxy. Langfuse upserts a
trace by id, so every turn of a session collapsed into one growing trace and
the Sessions view showed "Total traces: 1"
Detect that aliasing in the Langfuse callback from the request headers the
callback already receives, and use litellm_call_id as the trace id for those
calls. session_id still carries the header value, so the turns stay grouped
under one session. An explicit x-litellm-trace-id, langfuse_trace_id, or
langfuse_existing_trace_id keeps its trace id, including when the caller sets
it to the same value as the session id
Co-authored-by: jesus <jesus@berri.ai>
* test(langfuse): cover direct-SDK callers without proxy request headers
* fix(langfuse): preserve session trace provenance
---------
Co-authored-by: jesus <jesus@berri.ai>
A team-scoped auto-router is stored under an internal
model_name_{team_id}_{uuid} with the caller-facing name in
model_info.team_public_model_name, and the four pre-routing strategy
registries key on that internal name. A team key asks for the public name,
so the strategy lookup missed, the team early-resolve exit handed back the
marker deployment itself, and every call 400'd with "Unmapped LLM provider".
The strategy lookup now resolves the requested name through the same
team-first, then global, then admin-across-teams deployment resolution the
deployment path uses, and looks the registries up under the model_name of
whatever that resolves to. Both exits of _common_checks_available_deployment
drop strategy markers through one helper, so a marker-only resolution is
rejected as uncallable on every path. The request team id has one reader.
Resolves LIT-7363
Claude-Session: https://claude.ai/code/session_01NU97S7d2FUDDvTk59k53Wp
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
The http port serves /metrics/ behind virtual-key auth, so a ServiceMonitor
pointed at it only collects 401s and the RPM/TPM HPA metrics never appear
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* fix(proxy): accept non-string callback vars in default_team_settings
A YAML boolean such as turn_off_message_logging: true in a
default_team_settings block failed TeamCallbackMetadata's str-only
callback_vars validation and errored the request before any callback
ran. Stringify the value the same way AddTeamCallback does.
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* test(proxy): drop docstring from default_team_settings bool regression test
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* test(proxy): move default_team_settings bool regression test to mapped pre_call_utils suite
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
---------
Co-authored-by: yucheng <yucheng@berri.ai>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
LiteLLM_JWTKeyMapping_token_fkey was created ON DELETE RESTRICT, so deleting
a virtual key that a JWT mapping pointed at failed with a foreign key
violation on every deletion path (/key/delete, Admin UI, alias delete,
team and user cascades). Declaring onDelete: Cascade on the relation lets
the database clean the mapping up uniformly, so the next JWT call from that
identity re-registers against the newly created key.
Rebase of #33703 onto current staging.
Claude-Session: https://claude.ai/code/session_011Tn3657NkV6ojLqewL64Kb
The Rust token counter now mirrors _count_input_tokens key precedence
(messages, prompt, input, query/documents) so every LLM route that goes
through budget reservation gets the GIL-free count, not only /v1/messages
and /v1/chat/completions. Objects are serialised like json.dumps before
tokenizing; floats and unknown shapes still decline to Python. The body
model is optional so route-selected models can be matched by the caller
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
When a provider returns choices as null, an object, a string or a number, the
converter said the response had no 'choices' even though the key was present in
the raw keys it listed. A shared message now keeps the old wording for a missing
key and names the offending type otherwise.
The cached-stream regression test also pins the chunk count so a leaked extra
chunk fails it.
* fix(proxy): keep a body litellm_session_id in SpendLogs under missing_session_id omit
Under general_settings.missing_session_id: omit, apply_missing_session_id_policy now
mirrors a client-supplied top-level litellm_session_id into metadata.session_id when the
client did not set one there, so SpendLogs.session_id and Langfuse agree with the session
callbacks already report through StandardLoggingPayload.session_id
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* fix(proxy): keep client metadata.session_id ahead of body litellm_session_id on litellm_metadata routes
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
* test(proxy): drop docstrings from the missing_session_id omit regression tests
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
---------
Co-authored-by: yucheng <yucheng@berri.ai>
Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
Skipping the name write-back in either handler left every test green; a
guardrail that renames a tool call now has a regression test on both the
chat chunk path and the Anthropic SSE path
Preset #40341 pointed the Anthropic family REASONING tier at claude-fable-5-1,
but this test still hardcoded claude-opus-5, so the payload it saw no longer
matched. Rebase the assertion on ANTHROPIC_PRESET.complexity_router_config.tier_model_configs
so a preset refresh flows through instead of redding the suite on staging.
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: Krrish Dholakia <krrish-berri-2@users.noreply.github.com>
A stream cache hit on an entry stored with choices == [] indexed choices[0]
in the cached_response branch and failed with IndexError, so the streaming
converters' empty chunk had no working consumer. The branch now treats a
chunk without choices as empty and lets the wrapper close the stream with
its usual finish_reason stop chunk
The two test methods, the policy_engine fixture, and the two inner
stubs in TestBackgroundResponseRetrievalGovernance now carry full
parameter and return annotations, closing the Greptile thread that
94f9230d13 left open.
The Databricks chat transformation only parsed reasoning out of FMAPI-style
reasoning content blocks, so external models behind Databricks AI Gateway that
return the OpenAI-style top-level reasoning_content string lost it, both in the
final message and in every streamed delta. Fall back to the shared OpenAI
reasoning helper when no reasoning block exists, and keep the delta's own
reasoning_content when streaming.
Narrows the no-choices guard so a dict, string, or None still raises the APIError while an empty list passes through,
guards the non-stream Anthropic bridge against indexing an empty choices list, and repairs test_completion_missing_role,
whose raw-response mock was patched in as the create() callable itself so the handler only ever saw a MagicMock
