scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-09-17 23:51:30 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 66
49086 commits 2515 branches 1410 tags 3 GiB
Commit graph

49086 commits

This branch
This branch
All branches
Author SHA1 Message Date
yucheng
0d03ea3154 Merge remote-tracking branch 'origin/main' into litellm_agent365_mcp_guardrail
Some checks failed
LiteLLM Rust / rust-lint (push) Has been cancelled
Details
ai-gateway image / ai-gateway release image (push) Has been cancelled
Details
LiteLLM Rust / rust-test (push) Has been cancelled
Details
2026-09-15 00:30:12 +00:00
yucheng
4bfaf74edc refactor(guardrails): keep Agent 365 PR to the guardrail, move MCP changes to stacked PRs 
The listed-tool metadata and per-caller catalog cache move to a follow-up PR
stacked on this one, and the RFC 9728 sign-in challenge and discovery metadata
move to a second PR stacked on that. On its own the guardrail evaluates the
tool name and arguments and the caller presents its Entra token

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
 2026-09-15 00:29:58 +00:00
yucheng
b030164f73 Merge remote-tracking branch 'origin/main' into litellm_agent365_mcp_guardrail 2026-09-15 00:26:03 +00:00
yucheng-berri
b67137b67f
Merge pull request #40702 from BerriAI/litellm_model_armor_logging_only_streaming 
feat(model_armor): logging_only mode scans completed streams after delivery
 2026-09-14 17:23:32 -07:00
Yassin Kortam
d1fc231545
Merge pull request #41145 from BerriAI/litellm_health_check_skip_save_on_failed_read_main 2026-09-14 16:48:47 -07:00
yucheng-berri
99245f9323
Merge pull request #41131 from BerriAI/litellm_prompt_security_created_status 
fix(prompt_security): keep polling file sanitization through non-terminal statuses
 2026-09-14 16:43:37 -07:00
tin-berri
c626ff098b
Merge pull request #40877 from BerriAI/litellm_lit7658_cache_cost_v0_fresh 
feat(proxy): predict prompt-cache costs across deployments
 2026-09-14 16:25:41 -07:00
yuneng-jiang
9e485b0d5c
Merge pull request #41142 from BerriAI/litellm_codeql_import_noise 
ci(codeql): exclude noisy Python quality queries
 2026-09-14 16:25:38 -07:00
kerry-berri
9481556b5d
Merge pull request #40955 from BerriAI/litellm-providers/price-sync
Some checks are pending
LiteLLM Rust / rust-test (push) Waiting to run
Details
Unit Tests: Documentation Validation / documentation (push) Waiting to run
Details
Unit Tests: Proxy DB Operations / assert-shard-coverage (push) Waiting to run
Details
Unit Tests: Proxy DB Operations / auth-checks (push) Blocked by required conditions
Details
Unit Tests: Proxy DB Operations / budgets (push) Blocked by required conditions
Details
Unit Tests: Proxy DB Operations / custom-logging (push) Blocked by required conditions
Details
Unit Tests: Proxy DB Operations / db-and-spend (push) Blocked by required conditions
Details
Unit Tests: Proxy DB Operations / endpoints-and-responses (push) Blocked by required conditions
Details
Unit Tests: Proxy DB Operations / guardrails-hooks (push) Blocked by required conditions
Details
Unit Tests: Proxy DB Operations / jwt-and-keys (push) Blocked by required conditions
Details
Unit Tests: Proxy DB Operations / key-generation (push) Blocked by required conditions
Details
Unit Tests: Proxy DB Operations / logging-misc (push) Blocked by required conditions
Details
Unit Tests: Proxy DB Operations / proxy-runtime (push) Blocked by required conditions
Details
Unit Tests: Proxy DB Operations / proxy-server-core (push) Blocked by required conditions
Details
Unit Tests: Proxy DB Operations / proxy-utils (push) Blocked by required conditions
Details
Unit Tests / caching-local (push) Waiting to run
Details
Unit Tests / core-utils (push) Waiting to run
Details
Unit Tests / enterprise-package (push) Waiting to run
Details
Unit Tests / enterprise-routing (push) Waiting to run
Details
Unit Tests / integrations (push) Waiting to run
Details
Unit Tests / All Other Providers (push) Waiting to run
Details
Unit Tests / Vertex AI (push) Waiting to run
Details
Unit Tests / misc (push) Waiting to run
Details
Unit Tests / proxy-auth (push) Waiting to run
Details
Unit Tests / proxy-extras (push) Waiting to run
Details
Unit Tests / proxy-infra (push) Waiting to run
Details
Unit Tests / proxy-server (push) Waiting to run
Details
Unit Tests / responses-caching-types (push) Waiting to run
Details
GitHub Actions Security Analysis / zizmor (push) Waiting to run
Details
Unit Tests / proxy-endpoints (push) Waiting to run
Details 
chore(prices): sync Vertex AI prices: 14 models
 2026-09-14 16:16:07 -07:00
Yassin Kortam
9cbd239488
Merge pull request #41056 from BerriAI/litellm_moderation_call_id_header 
fix(proxy): resolve x-litellm-call-id from response metadata when routes omit call_id
 2026-09-14 16:14:13 -07:00
Yuneng Jiang
15fdc5f0f0
ci(codeql): retain Python log-injection coverage 2026-09-14 16:09:27 -07:00
yassin
c6a36fb8b6 fix(health): skip background health check DB writes when the latest-row read fails 
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
(cherry picked from commit 6d2c4899b0)
 2026-09-14 23:03:30 +00:00
yassin
ece2bbfed7 fix(proxy): resolve x-litellm-call-id from response metadata when routes omit call_id 
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
 2026-09-14 23:03:27 +00:00
Yuneng Jiang
270ce9d8ca
ci(codeql): exclude remaining audited Python query categories 2026-09-14 16:01:12 -07:00
yucheng
f235c8b3a2 fix(agent_365): keep sign-in discovery for servers forwarding an API-key header
Some checks are pending
ai-gateway image / ai-gateway release image (push) Waiting to run
Details
LiteLLM Rust / rust-lint (push) Waiting to run
Details
LiteLLM Rust / rust-test (push) Waiting to run
Details 
A forwarded x-api-key, api-key or apikey header is the upstream's own credential and leaves the caller's top-level Authorization with the gateway, so Agent 365 discovery and the connect-time challenge must not be suppressed for those servers. Only modes or forwarded headers that consume top-level Authorization opt a server out.

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
 2026-09-14 22:59:26 +00:00
yuneng-jiang
63386d6cc6
Merge pull request #41070 from BerriAI/litellm_integration_management 
test: cover persisted updates and warmed authorization policies
 2026-09-14 15:57:32 -07:00
Yuneng Jiang
97e9748048
ci(codeql): exclude noisy Python import-cycle queries 2026-09-14 15:57:24 -07:00
Yassin Kortam
de55e22899
Merge pull request #41141 from BerriAI/litellm_lit7694_forwarded_headers_body_leak 
fix(openai): keep extra_headers out of the chat request body on the httpx handler path
 2026-09-14 15:52:31 -07:00
Yassin Kortam
1f46e58494
Merge pull request #41018 from BerriAI/litellm_fix_model_alias_map_leak 
fix(utils): keep litellm params out of provider request bodies
 2026-09-14 15:51:23 -07:00
yuneng-jiang
c57957f45d
Merge pull request #41066 from BerriAI/litellm_integration_foundation 
test: add dedicated CircleCI integration contract foundation
 2026-09-14 15:50:22 -07:00
kerry
b25b6eb5a4 test(cost_calc): pin vertex_ai gemini-3.5-flash-lite priority cache reads at the published 5.4e-08 
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
 2026-09-14 22:48:59 +00:00
yassin
e2e1d36804 fix(openai): keep extra_headers out of the chat request body on the httpx handler path 
Forwarded client headers on bridged /v1/responses calls were serialized into the
OpenAI JSON body as extra_headers when EXPERIMENTAL_OPENAI_BASE_LLM_HTTP_HANDLER
was set, and OpenAI rejected the request with unknown_parameter. The headers are
already merged into the outgoing HTTP headers, so only set the SDK-style
optional param on the SDK client path

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
 2026-09-14 22:27:35 +00:00
yucheng
6b9990e068 fix(agent_365): treat Entra AADSTS50027xx invalid_client as a caller assertion failure 
Entra reports a forged or garbled On-Behalf-Of assertion as invalid_client, the
same OAuth error it uses for a bad gateway client secret. Read the AADSTS
error_codes so a malformed assertion sends the caller back to sign in at connect
and at the tool call instead of reporting the gateway's credentials as broken.

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
 2026-09-14 22:12:12 +00:00
Devin AI
6f9915b650 fix(utils): keep litellm params out of provider request bodies 
get_litellm_params() keys such as model_alias_map, completion_call_id and
custom_prompt_dict were missing from all_litellm_params, so when a bridge
forwarded litellm_params as completion kwargs they were swept into extra_body
and OpenAI rejected the call with "Unknown parameter: 'model_alias_map'".

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
 2026-09-14 22:07:27 +00:00
ryan-crabbe-berri
7fd541efb9
Merge pull request #41048 from HUAHAODIA/litellm_ratchet_strict_rules 
chore(lint): graduate 12 rules from the strict-gate ratchet
 2026-09-14 15:06:16 -07:00
ryan-crabbe-berri
7b6d223105
Merge pull request #40196 from BerriAI/litellm_team_admin_all_proxy_models 
fix(ui): let team admins grant a team all proxy models
 2026-09-14 15:05:01 -07:00
yucheng
0e27a523e0 fix(mcp): probe Entra at connect, challenge only default-on Agent 365 guardrails, key listed tools by signer 
Connect-time sign-in challenge now asks Entra to exchange the presented assertion instead of only checking
its compact-JWS shape, so expired, wrong-audience or forged bearers get the RFC 9728 challenge while gateway
credential and provider failures still surface on the tool call. Only default_on guardrails the caller has not
opted out of advertise or challenge, since anonymous metadata cannot see key-selected guardrails. Servers whose
Authorization is minted per caller by MCPJWTSigner list tools per caller instead of sharing one cache slot

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
 2026-09-14 22:04:02 +00:00
ryan-crabbe-berri
51926c2e79
Merge pull request #41023 from BerriAI/litellm_ui_move_metadata_tags_to_tags_field 
fix(ui): move tags typed into key metadata JSON into the Tags field
 2026-09-14 15:03:54 -07:00
yucheng
3a3075b8a2 fix(prompt_security): poll only on queued statuses, keep 500 for terminal or missing status 
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
 2026-09-14 22:01:39 +00:00
berriai-litellm-provider-info-sync[bot]
c3f8c07c43
chore(prices): sync Vertex AI prices: 14 models 
vertex_ai/deep-research-pro-preview-12-2025: cache_read_input_token_cost
vertex_ai/gemini-2.5-flash-image: input_cost_per_token_flex, output_cost_per_token_flex, input_cost_per_token_batches, input_cost_per_token_priority, output_cost_per_token_batches
vertex_ai/gemini-3-flash-preview: input_cost_per_token_flex, output_cost_per_token_flex, input_cost_per_token_batches, output_cost_per_token_batches, cache_read_input_token_cost_flex, cache_read_input_audio_token_cost
vertex_ai/gemini-3-pro-image: input_cost_per_token_flex, output_cost_per_token_flex, cache_read_input_token_cost, input_cost_per_token_priority, output_cost_per_token_priority, cache_read_input_token_cost_flex, cache_read_input_token_cost_priority, input_cost_per_token_above_200k_tokens, output_cost_per_token_above_200k_tokens, cache_read_input_token_cost_above_200k_tokens, input_cost_per_token_above_200k_tokens_priority, output_cost_per_token_above_200k_tokens_priority, cache_read_input_token_cost_above_200k_tokens_priority
vertex_ai/gemini-3.1-flash-image: input_cost_per_token_flex, output_cost_per_token_flex, cache_read_input_token_cost, input_cost_per_token_batches, output_cost_per_token_batches, cache_read_input_token_cost_flex
vertex_ai/gemini-3.1-flash-lite: cache_read_input_audio_token_cost
vertex_ai/gemini-3.1-flash-lite-image: input_cost_per_token_flex, output_cost_per_token_flex, cache_read_input_token_cost_flex
vertex_ai/gemini-3.1-pro-preview: input_cost_per_token_flex, output_cost_per_token_flex, cache_read_input_token_cost_flex
vertex_ai/gemini-3.5-flash: 
vertex_ai/gemini-3.5-flash-lite: cache_read_input_token_cost_priority
vertex_ai/gemini-3.6-flash: 
vertex_ai/gemini-3.7-flash: 
vertex_ai/gemini-3.8-flash: 
vertex_ai/gemini-embedding-2:
 2026-09-14 22:00:57 +00:00
Yassin Kortam
e766277846
Merge pull request #40991 from BerriAI/litellm_team_model_cooldown_siblings 
fix(router): cool down team deployments on 429 when a sibling serves the same public model
 2026-09-14 14:57:07 -07:00
Yassin Kortam
3e417fa6e0
Merge pull request #41113 from BerriAI/litellm_realtime_release_max_parallel_slot 
fix(proxy): release max_parallel_requests slot when a realtime session ends without LLM callbacks
 2026-09-14 14:56:11 -07:00
ryan-crabbe-berri
f9ba42860c fix(azure_ai): keep api_version as a property alongside the new getter 
The PLR0206 fix renamed the property to get_api_version and dropped it.
AzureFoundryModelInfo().api_version does resolve today, so removing it
breaks external callers. Restore it as a zero-arg property delegating to
the staticmethod, which satisfies PLR0206 either way.
 2026-09-14 14:56:07 -07:00
ryan
52a3b0ede7 fix(ui): trim and drop blank tags moved out of key metadata JSON 
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
 2026-09-14 21:50:25 +00:00
yujonglee
9bbb869b13
Merge pull request #41120 from BerriAI/litellm_trace_profiler_threads 
refactor(harness): expand independent trace coverage
 2026-09-14 14:50:18 -07:00
Devin AI
e4ebeae800 fix(prices): add text output rate to Vertex TTS entries, sync gemini-embedding-2 alias 
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
 2026-09-14 21:49:15 +00:00
Yassin Kortam
24bfd5fba1
Merge pull request #41102 from BerriAI/litellm_team_membership_once_main 
fix(auth): load team membership once per request and skip prisma on an L1 hit
 2026-09-14 14:46:06 -07:00
yassin
fbc1011d27 fix(bedrock): end the realtime session when the client disconnects instead of waiting for Nova Sonic 
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
 2026-09-14 21:42:42 +00:00
Devin AI
f45e20e6c2 fix(prices): mark Vertex Gemini TTS entries as audio_speech 
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
 2026-09-14 21:41:43 +00:00
kerry
abbf5aae20 chore(prices): flag gpt-5.5-cyber as a reasoning model 
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
 2026-09-14 21:41:10 +00:00
ryan
df87a49f8b fix(ui): offer no models when an org team's ceiling never arrives 
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
 2026-09-14 21:35:55 +00:00
yassin
b4d0f4ad26 refactor(realtime): move session ownership marker keys into constants 
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
 2026-09-14 21:33:48 +00:00
yucheng
5355cbe235 fix(mcp): keep Agent 365 evaluation faithful to each caller's tool call 
Key the listed-tool cache by every request input that changes the upstream catalog (caller identity, per-server auth header, forwarded headers, header-driven stdio env, relayed bearer) so one caller's tool metadata never reaches Agent 365 for another. Advertise Entra in anonymous protected-resource discovery only for default_on guardrails. Check the key's server grant before issuing the Agent 365 sign-in challenge so a key without access gets the grant's 403. Group conversationId by the MCP session id, falling back to the call id. Run Agent 365 in the parallel block-only phase so it judges the arguments the sequential guardrails hand upstream. Apply tool_name_to_description to local-registry tool metadata sent to Agent 365

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
 2026-09-14 21:28:18 +00:00
berriai-litellm-provider-info-sync[bot]
93e65995d6
chore(prices): sync Vertex AI prices: 4 models 
gemini-2.5-flash: cache_read_input_audio_token_cost
gemini-2.5-flash-lite: cache_read_input_audio_token_cost
gemini-3-flash-preview: cache_read_input_audio_token_cost
gemini-3.1-flash-lite: cache_read_input_audio_token_cost
 2026-09-14 21:26:02 +00:00
kerry
ddf13e9505 Merge remote-tracking branch 'origin/main' into litellm-providers/price-sync 2026-09-14 21:21:55 +00:00
kerry-berri
c6e4c5582d
Merge pull request #41093 from BerriAI/litellm_fallback_backfill_opt_in_main 
feat(model_info): provider-scoped fill_missing_for_providers backfill from fallback generalization rules
 2026-09-14 14:20:36 -07:00
yassin
d2342f06ce fix(bedrock): stamp the realtime success ownership marker when Nova Sonic spend is logged 
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
 2026-09-14 21:13:29 +00:00
yassin
db8dfe93a5 fix(auth): wait for the in-flight membership load before evicting its cache key on invalidation 
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
 2026-09-14 21:10:43 +00:00
yassin
2441e8a2a9 Merge remote-tracking branch 'origin/main' into litellm_realtime_release_max_parallel_slot 2026-09-14 21:03:28 +00:00
yucheng
fb60f80c80 fix(prompt_security): keep polling file sanitization through non-terminal statuses 
Prompt Security reports a queued sanitization job as status "created" before
it moves to "in progress" and "done". The poller treated anything other than
those two known strings as an error and returned HTTP 500 on the first poll, so
every image or file request through the guardrail failed while the vendor job was
still queued.

Only "done" is terminal now. Every other status is logged and polled again until
max_poll_attempts or the outer file_sanitization_timeout, after which the existing
fail-open or fail-closed (408) policy applies.

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
 2026-09-14 21:03:02 +00:00