BerriAI/litellm v1.111.0-dev.1 on GitHub

Verify Docker Image Signature

All LiteLLM Docker images are signed with cosign. This release is signed with the key introduced in commit e82a3c6.

Verify using the pinned commit hash (recommended):

A commit hash is cryptographically immutable, so this is the strongest way to ensure you are using the published signing key:

cosign verify \
  --key https://raw.githubusercontent.com/BerriAI/litellm/e82a3c646908ba0a6ee0beddc4310d7a6cac4065/cosign.pub \
  ghcr.io/berriai/litellm:v1.111.0-dev.1

Verify using the release tag (convenience):

Tags are protected in this repository and resolve to the same key. This option is easier to read but relies on tag protection rules:

cosign verify \
  --key https://raw.githubusercontent.com/BerriAI/litellm/v1.111.0-dev.1/cosign.pub \
  ghcr.io/berriai/litellm:v1.111.0-dev.1

Expected output:

The following checks were performed on each of these signatures:
  - The cosign claims were validated
  - The signatures were verified against the specified public key

What's Changed

fix(completion): accept a tool_choice that forces an OpenAI hosted tool by @kerry-berri in #50578
fix(accounting): stop delivery metrics from scanning retained receipts on every flush by @joshua-berri in #50565
fix(key/info): return 400 instead of 500 when no key is provided by @apex-mochen in #43577
fix(gemini): add the Flex above-200k cache read price to gemini-3.1-pro-preview rows by @devin-ai-integration[bot] in #50607
test(observability): match the full card number instead of a 4111 substring that timestamps can contain by @devin-ai-integration[bot] in #50612
fix(proxy): align ownership and scope handling by @mateo-berri in #50600
feat(proxy): total_timeout bounds a whole non-streaming request by @joshua-berri in #50496
fix(cli): end the local login when lite logout cannot check the OS keychain by @zoveki in #50098
fix(ui): label every budget reset time in the Admin UI with UTC by @zoveki in #50591
fix(proxy)!: price expanded wildcard /model/info rows as the model their calls route to by @paveln83 in #49128
fix(bedrock): derive the web identity session policy per AWS caller by @tavren27 in #50592
perf(router): skip unused deployment reads in the callback filter step and cache anthropic capability probes by @narmstrong-berri in #50641
test(integration): follow cached prepared statements in the held statement relay by @devin-ai-integration[bot] in #50651
feat(proxy): add OpenSearch index pass-through with per-key read/write grants by @quietjules in #50573
test(e2e): repin the Together reasoning replay backend to MiniMax-M3.1-Flash by @yuneng-berri in #50601
fix(proxy): build plugin session claims from FIPS provider services in fips_mode by @yucheng-berri in #50544
feat(anthropic): add claude-fable-5-2 and its Claude Platform on AWS mirror by @devin-ai-integration[bot] in #50715
feat(cost-map): rolling registry update (Anthropic and Bedrock Fable 5.2, Azure realtime cache, Vertex, Friendli) by @devin-ai-integration[bot] in #50707
fix(proxy): show committed spend in the remaining budget gauges by @zoveki in #49199
feat(azure): add azure_ai/claude-fable-5-2 and claude-fable-5-2-fast from the Foundry Claude page by @devin-ai-integration[bot] in #50723
build(ui): fail lint on hardcoded English in dashboard JSX by @yuneng-berri in #50593
fix(prompt_templates): keep moved private helpers importable from common_utils by @zoveki in #50721
perf(router): look up host step answers by id instead of scanning every candidate by @narmstrong-berri in #50704
feat(cost-map): add Claude Fable 5.2 on Vertex AI and repoint claude-fable-latest by @kerry-berri in #50713
test(integration): cover claude-fable-5-2 capability flags and prompt-cache pricing on the scripted Anthropic upstream by @kerry-berri in #50722
fix(cost)!: bill transcriptions by the seconds the provider reports by @ryan-crabbe-berri in #50727
fix(logging): bill a refused response the router falls back from by @joshua-berri in #50735
fix(audio): measure duration for PathLike multipart tuple contents by @raunakpandey1 in #49806
feat(auth): grant access groups through principal attribute rules by @krrish-berri-2 in #50455
fix(minimax): forward reasoning_effort instead of dropping it by @jeevan1098 in #49583
fix(ollama_chat): give each streamed tool call its own index by @tedswinyar in #49183
fix(redis): accept a bare service account email for GCP Memorystore IAM by @yujonglee-berri in #50676
fix(rust): install the rustls crypto provider at the Python host boundary by @yujonglee-berri in #50502
feat(cost-map): add GovCloud regional rows for Bedrock Claude Fable 5.2 and Mythos 5.2 by @devin-ai-integration[bot] in #50746
fix(bedrock): answer 400, not 500, for malformed count-tokens requests by @paveln83 in #50708
feat(router): let deployments declare a shared encrypted content boundary by @joshua-berri in #50750
test(integration): cover claude-fable-5-2 capability flags and prompt-cache pricing on Vertex AI and Bedrock Converse by @kerry-berri in #50742
fix(proxy): fail-closed budget enforcement returns 403 naming /team/member_add for a team user with no membership row by @kerry-berri in #50603
fix(complexity_router): keep Claude Code subagent and side-request traffic off other agents' tier pins by @joshua-berri in #50741
test(audio): bill reported transcription seconds for measurable uploads by @devin-ai-integration[bot] in #50767
chore(cost-map): add openai gpt-6-cyber long context prices from the pricing page by @devin-ai-integration[bot] in #50770
feat(proxy): let a key's logging entry reference a proxy credential by @jesus-berri in #50572
fix(messages): let proxy /v1/messages requests run on the rust path by @narmstrong-berri in #50335
fix(messages): log native messages success once, after the proxy releases it by @narmstrong-berri in #50420
fix(messages): log native messages request fields like python by @narmstrong-berri in #50421
fix(messages): pass non-message bodies through and refuse non-string headers as bad requests by @narmstrong-berri in #50422
fix(rust): run native async calls through python's async wrapper steps by @narmstrong-berri in #50423
fix(messages): raise native messages failures as python's errors by @narmstrong-berri in #50424
fix(messages): bill a native stream the client abandoned like python by @narmstrong-berri in #50425
fix(messages): write the native unary cache entry after the router stamps it like python by @narmstrong-berri in #50426
test(integration): assert response ids stay scoped to their minting key by @devin-ai-integration[bot] in #50703
chore(cost-map): add github_copilot/claude-fable-5.2 from the Copilot models-and-pricing table by @devin-ai-integration[bot] in #50775
test(integration): assert team model visibility follows caller scope by @devin-ai-integration[bot] in #50772
fix(cost-map): price wandb DeepSeek-V4.1-Flash-1017 from the CoreWeave Forge listing by @devin-ai-integration[bot] in #50776
fix(proxy): write SpendLogs failure rows for config-defined pass-through endpoints by @devin-ai-integration[bot] in #41831
test(integration): repair tests that merged changes left stale by @yuneng-berri in #50782
fix(router): merge default metadata under the request's instead of replacing it by @paveln83 in #50700
fix(anthropic-messages): apply additional_drop_params before the unsupported param checks by @tavren27 in #50753
fix(router): honor request fallbacks in the mid-stream continuation check by @narmstrong-berri in #50260
fix(router): let a retry leave a pinned deployment after a retryable failure by @joshua-berri in #50769
test(integration): drop the empty-string n case from the Gemini long decode test by @yuneng-berri in #50796
feat(rust): add litellm-s3 crate, AwsCredentialsProvider and cost map fetch by @yujonglee-berri in #50547
refactor(model-cost): move the cost map inputs, fetch and register_model overlay into Rust by @yujonglee-berri in #50501
fix(prometheus): record litellm_overhead_latency_metric for streams, stamp plus the proxy's first-chunk work by @kerry-berri in #49920
fix(accounting): keep delivering spend when another worker's maintenance lock is orphaned by @mateo-berri in #50774
fix(sso): read the SSO client secret before the refresh-token DB check by @mateo-berri in #50773
test(integration): expect the 400 for a content-less leading system on Mantle count_tokens after #50708 by @devin-ai-integration[bot] in #50820
fix(proxy): serve Codex its own Bedrock catalog entries for Bedrock deployments by @ryan-crabbe-berri in #50743
feat(usage): split customer spend by the key's current team by @kenneth-berri in #50764
fix(proxy): list images in the Codex catalog only for models known to read them by @ryan-crabbe-berri in #50738
fix(hosted_vllm): omit empty tools list from chat requests by @krrish-berri-2 in #50812
feat(proxy): log keys and teams still on the deprecated empty models defaults at boot by @paveln83 in #50720
fix(router): count a Responses WebSocket turn's tokens before the next turn is admitted by @norkelbyte in #50730
chore(cost-map): add the openai gpt-image-2 shutdown date from the deprecations page by @devin-ai-integration[bot] in #50829
chore(cost-map): add together_ai/elorian-labs/glimpse-1.0 from the Together models API by @devin-ai-integration[bot] in #50830
fix(proxy)!: return managed file ids from rag ingest and use team credentials on vector store file routes by @tavren27 in #50459
test(decisions): ignore Ollama model-info calls in the decoy check by @moe-berri in #50827
fix(complexity_router): hold the default model for a user_turn ask whose classifier failed by @joshua-berri in #50793
test: update the spend tests that still expected the pre-ledger accounting queues by @yuneng-berri in #50814
feat(budgets): add per-model tpm and rpm limit maps to budgets and customers by @kerry-berri in #50797
fix(ui): keep the credential action menu open when the models page re-renders by @devin-ai-integration[bot] in #50845
fix(transcription): stop billing a FLAC with no sample count as 576 trillion seconds by @kerry-berri in #50842
fix: make scheduled key, user, and team spend resets idempotent by @mateo-berri in #50298
feat(cost-map): add meta/muse-ember-1.0 with long-context rates from the Meta pricing page by @kerry-berri in #50847
test(accounting): assert spend is delivered under another leader's maintenance lock by @devin-ai-integration[bot] in #50843
fix(proxy): send the whole Codex model catalog to Codex 0.166.0-alpha.1 and newer by @ryan-crabbe-berri in #50736
fix(proxy): let the OS assign Prisma engine ports by @devin-ai-integration[bot] in #50850
fix(ui): keep the request log model cell to one model by @Bytechoreographer in #49619
test(integration): empty the shared client cache before the tier pin batch test by @yuneng-berri in #50854
feat(cost-map): rolling registry update (claude-mythos-5-2, Azure Mythos 5.2, Friendli GLM-5.1 date, GovCloud Fable 5.2 token counting) by @devin-ai-integration[bot] in #50777
fix(proxy): bill an OpenAI pass-through background response once across repeated reads by @tavren27 in #50841
feat(cost-map): add gemini/gemini-4-carbon from the Gemini API pricing page by @devin-ai-integration[bot] in #50931
fix(redis): free each request's Redis batch scope once the request ends by @emerzon in #50751
fix(openai): download PDF URLs off the event loop and transform async chat messages once by @norkelbyte in #49176
feat(cost-map): add vertex_ai/gemini-4-carbon and gemini-4-carbon-cyber-permissive from the Vertex pricing page by @devin-ai-integration[bot] in #50942
refactor: clean up fresh tech debt in stream cost flag, masked httpx error and encrypted reasoning helpers by @mateo-berri in #50617
fix(proxy): open the Logs drawer by request_id without scanning for litellm_call_id by @tavren27 in #50789
fix(cost-map): give the Gemini 4 Carbon rows the shared Gemini 3 search free tier by @kerry-berri in #50951
perf(streaming): reuse the upstream choice for plain litellm HTTP-handler stream chunks by @quietjules in #50786
fix(responses): replay Claude tool search on follow-up turns by @zoveki in #50821
feat(rate-limit): enforce per-customer per-model tpm and rpm limits in the v3 limiter by @kerry-berri in #50857
test(cost): sync vertex 3.1 pro price fields and bedrock converse usage shapes by @devin-ai-integration[bot] in #50956
fix(decisions): keep box questions on a deployment when a same-model sibling turns them off by @tavren27 in #50724
feat: add claude mythos 5.2 to vertex ai cost map by @kerry-berri in #50940
fix(router): hold the registry lock across router-core writes and mirror reads by @narmstrong-berri in #50933
test(budgets): expect per-model tpm and rpm limits in budget info by @devin-ai-integration[bot] in #50967
feat(ui): customer portal (beta) surfacing the existing customer endpoints and customer usage by @kerry-berri in #50807
feat(customers): reject unknown fields on /customer/new and /customer/update instead of dropping them by @kerry-berri in #50965
fix(proxy): send a prompt-only call_endpoint token count to every provider count API by @paveln83 in #50783
test(integration): cap the OAuth metadata lookup in the mixed capabilities subscription case by @yuneng-berri in #50867
fix(mcp): accept comma-separated vendor scope grants on the connect check by @shrey-berri in #50717
fix(cost-map): give the Gemini 4 Argon rows the shared Gemini 3 search free tier by @kerry-berri in #50959
test(http2): drain client traffic after sending GOAWAY by @devin-ai-integration[bot] in #50972
fix(responses): keep unset-effort gpt-5.4/5.5 tool calls on chat completions by @devin-ai-integration[bot] in #50968
fix(cost-map): mark the azure_ai Claude rows that bill web search as supporting it by @kerry-berri in #50960
test(integration): recognize the model list's lock in the held write helper by @yuneng-berri in #50974
test(anthropic): scope structured output captures to inference routes by @devin-ai-integration[bot] in #50983
feat(proxy): let teams use their own allowed S3 buckets for batch files by @mrinal-berri in #50571
fix(router): let an explicit routing tag outrank the encrypted content pin by @joshua-berri in #50975
feat(cost): bill batch cached image input tokens for OpenAI image models by @devin-ai-integration[bot] in #50997
fix: audit registry metadata for Together, Mistral, Fireworks, Azure AI and Copilot by @devin-ai-integration[bot] in #51003
fix(cost-map): sync together_ai together/Tev1-4B-experimental input price with /v1/models by @devin-ai-integration[bot] in #51013
chore(cost-map): add github_copilot/gemini-4-carbon from the Copilot changelog launch post by @devin-ai-integration[bot] in #51012
fix(proxy)!: reject client-supplied Vertex location and project unless the admin opts in by @mateo-berri in #47028
fix(router): sync responses() applies the Router's num_retries and fallbacks by @joshua-berri in #51011
feat(spend): report guardrail spend separately in daily activity by @joshua-berri in #51010
fix(tool_choice): accept the Responses API's flat allowed_tools layout on chat completions by @mateo-berri in #50467
test(integration): keep owned test peers off ephemeral ports by @devin-ai-integration[bot] in #51019
fix(bedrock): forward x-pass- headers on the /bedrock/model pass-through routes by @norkelbyte in #50728
fix(proxy): send the streamed Responses fallback metadata as a typed Responses event by @paveln83 in #50804
fix(proxy): recheck missing users when resolving session grants by @devin-ai-integration[bot] in #51030
feat(cli): connect a per-user GitHub Copilot credential from the command line by @ryan-crabbe-berri in #50863
fix(responses): keep the namespace == name shape for deferred top-level tools by @kerry-berri in #50602
chore(cost-map): add fal_ai image and video rows priced from the fal.ai model pages (first fal-ai sync) by @devin-ai-integration[bot] in #51035
fix(vertex_ai): bill gemini image edits by token usage by @devin-ai-integration[bot] in #51038
feat(search): add Microsoft Web IQ provider by @giladk1221 in #50883
fix(otel): serialize pydantic Message objects in langfuse_otel, weave_otel and langtrace by @taxfree-python in #46845
perf(spend): narrow user and team spend log reads to the owner's api_keys so the (api_key, startTime) index is used by @kerry-berri in #51026
fix(mcp): challenge a single-server listing whose credential the upstream rejected by @shrey-berri in #50389
fix(ci): install tzdata so zoneinfo tests pass on runners without legacy tz aliases by @mateo-berri in #51057
fix(guardrails/rubrik): moderate text-completion prompts and forward rac_agent_id by @seph-barker in #49912
fix(model-prices): use current Hy4 promotional rates by @moe-berri in #51072
fix(proxy): allow management_routes keys to call PATCH /model/{model_id}/update by @nurlanmnn in #50979
fix(proxy): drop the beta label from the /v1/messages swagger tags by @mateo-berri in #51046
fix(router): re-check and reserve per-model rate limits on fallback targets by @kerry-berri in #50994
chore(cost-map): add 61 fal_ai image and video rows priced from the fal.ai model pages (second fal-ai sync) by @devin-ai-integration[bot] in #51077
feat(auto_router): report a session's whole spend and name why savings stay unestimated by @joshua-berri in #51066
chore(pricing): sync replicate model prices from replicate.com model pages by @devin-ai-integration[bot] in #51067
chore(databricks): add claude-fable-5-2, date glm-5-2 retirement and fix llama-3-1-8b context from the supported-models page by @devin-ai-integration[bot] in #51076
refactor(rust_bridge): route every Rust-to-Python call through host modules by @yujonglee-berri in #50822
fix(spend): write a background response's settling read in-process when the spend collector is on by @tavren27 in #50957
fix(proxy): judge the zero-cost budget waiver by the model a key or team alias serves by @joshua-berri in #51058
feat(docker)!: one shipped image, one entrypoint beside the Dockerfile, one compose file by @devin-ai-integration[bot] in #43332
feat(anthropic): forward message threads on native /v1/messages by @paveln83 in #50976
fix(router): send managed batch uploads to batch deployments when a chat deployment has a lower order by @mika-lt in #50971
fix(rate-limit): settle a fallback's tokens on the model group that served it by @kerry-berri in #51082
fix(proxy): keep the upgrade notice off the request path and count area-labelled release notes by @kerry-berri in #51027
fix(vertex_ai): bill gemini image text, thinking and priority tokens at their own rates by @devin-ai-integration[bot] in #51017
test(spend): expect token billing for Vertex image edits with usage by @devin-ai-integration[bot] in #51109
fix(reasoning): log the GPT-5 reasoning_effort drop warning only when the provider gets no effort by @paveln83 in #50995
test(mcp): read the comma grant through real credential storage by @shrey-berri in #51065
fix(proxy): keep a request's cache opt-in when its key has cache metadata by @quietjules in #50999
feat(replicate): add google/gemini-4-carbon from the Replicate model page by @devin-ai-integration[bot] in #51125
fix(mcp)!: forward the LiteLLM key when the admin configures it by @joshua-berri in #50604
feat(azure): add azure_ai/mistral-large-4-latest from the Azure Models API, priced from Mistral's model page by @devin-ai-integration[bot] in #51047
ci(image-scan): build Rust with a fast ci cargo profile instead of fat-LTO release by @mateo-berri in #51075
test(unit): deflake db metrics spans and drain timeout tests, split proxy-infra shard by @mateo-berri in #51078
chore(cost-map): add 3 fal_ai rows priced from the fal pricing API and model page (third fal-ai sync) by @devin-ai-integration[bot] in #51126
fix(bedrock): govcloud eol dates from the lifecycle table, gpt-5.6 cyber govcloud mantle row and daybreak blue input limit by @devin-ai-integration[bot] in #51095
feat(helm): single chart with monolith mode and shared image, remove litellm-helm by @devin-ai-integration[bot] in #43330
feat(terraform): run every AWS and GCP workload from the one litellm image by @devin-ai-integration[bot] in #43358
fix(spend): keep empty-key rows in owner-scoped spend log reads by @mateo-berri in #51127
fix(proxy): list callable ids for partial aliases over a catch-all deployment by @quietjules in #50963
test(providers): preserve unspecified Gemini image output modality by @devin-ai-integration[bot] in #51149
test(spend): expect token billing for Vertex image edits by @devin-ai-integration[bot] in #51148
fix(gemini): batch and flex tier fields the -latest alias rows were missing by @devin-ai-integration[bot] in #51150
fix(spend): keep free response reads and cancels off the creation spend row by @devin-ai-integration[bot] in #51110
fix(proxy): add tzdata to the proxy extra so a legacy TZ doesn't stop startup by @satwik-bankapur in #49914
fix(tls)!: exclude RSA key exchange and ChaCha20 from the outbound cipher list in FIPS mode by @yucheng-berri in #50671
fix(vertex): drop floor-only retirement dates, veo 3.1 successor and gemini token limits from the Vertex model pages by @devin-ai-integration[bot] in #51180
docs(security): updated mailinglist and vdp in security.md by @ojensen-berri in #51199
chore(cost-map): rolling registry PR, fal_ai rows, databricks gemini-4-carbon, deepinfra and bedrock tier prices by @devin-ai-integration[bot] in #51193
fix(router): preserve held tiers for OpenCode subagent results by @joshua-berri in #51176
refactor: clean up fresh tech debt in rubrik, passthrough headers and plugin mode comments by @mateo-berri in #51163
test(security): expect the async release check on offline mode by @devin-ai-integration[bot] in #51213
fix(management): keep classifier deployments serving during reloads by @devin-ai-integration[bot] in #51111
chore(cost-map): add 24 runwayml models and correct seedance2 and eleven_multilingual_v2 rates by @devin-ai-integration[bot] in #51216
fix(azure): use supported API versions for Responses requests by @kenneth-berri in #51116
feat(proxy): count a rate-limit fallback in x-litellm-attempted-fallbacks and name it in x-litellm-fallback-reason by @kerry-berri in #51138
refactor(proxy): let the gunicorn server take a private listener by @yujonglee-berri in #50871
feat(proxy): supervise the opt-in Rust forwarding server by @yujonglee-berri in #49606
fix(streaming): report zero cost for explicitly free calls by @devin-ai-integration[bot] in #51093
feat(gemini): file-backed batch jobs on /v1/batches with output download and spend tracking by @kerry-berri in #51032
fix(router): map Gemini stream-opening errors by @joshua-berri in #51218
fix(auth): enforce model access on Google and image URL routes by @joshua-berri in #51226
test(integration): fix the baseline accounting and series cap flakes by @yuneng-berri in #51096
fix(proxy): let /team/delete transactions wait for a pooled connection by @yuneng-berri in #51099
fix(completion)!: answer 400 for a chat message without a role by @zoveki in #51124
fix(proxy): stream the guardrail block response when the block comes after an SSE keepalive ping by @paveln83 in #51114
feat(otel): let a team or key OTel v2 destination add its own request metadata keys by @mrinal-berri in #50813
refactor(typing): clear 1141 basedpyright Any errors across 182 files by @quietjules in #50711
fix(proxy): reset a deleted customer's admission spend counter so a re-created one is not refused by @joshua-berri in #51224
fix(proxy): answer x-litellm-call-id on every token count response by @tavren27 in #51037
feat(rate-limit): name the limit that rejected a request in a rate_limited_by header on the 429 by @kerry-berri in #51228
refactor(ui): remove the deprecated Learning Resources sidebar link by @yuneng-berri in #51251
feat(helm): add scheme and tlsConfig to the gateway ServiceMonitor by @mrinal-berri in #50759
fix(proxy): name the deployment in stream error alerts on chat, completions, responses and messages by @quietjules in #51034
refactor(ui): remove the deprecated API Reference page by @yuneng-berri in #51246
test(e2e): remove the secret manager suite by @yuneng-berri in #51241
fix(bedrock): forward the compact-2026-09-04 compaction param on Invoke /v1/messages by @kerry-berri in #51223
fix(fallback): preserve concurrent writes and serve saved configuration by @joshua-berri in #51245
chore: remove dead files from the repo root by @yassin-berriai in #50936
chore: remove cookbook and preserve live assets by @yassin-berriai in #50944
ci: remove CircleCI by @yassin-berriai in #50950
security(proxy): stop request body fields from conflicting with admin-set deployment fields by @shrey-berri in #51254
test(proxy): make the cancelled supervisor test deterministic by @yuneng-berri in #51258
fix(responses): keep reasoning item IDs consistent across streaming events by @joshua-berri in #51243
feat(guardrails): write guardrail changes to the audit log by @mrinal-berri in #51118
fix(llms-types): let the compact-2026-09-04 beta through on bedrock to match the headers config by @kerry-berri in #51272
feat(messages): serve anthropic messages on rust only by @narmstrong-berri in #51119
fix(health): log Test Connection spend with the mode's call type and model id by @kenneth-berri in #50987
feat(bedrock): search knowledge bases by image and return image hits by @zoveki in #51063
fix(fips): hash MCP advisory-lock keys with SHA-256 in fips_mode and compute the OCI body digest as a security hash by @yucheng-berri in #51165
test(proxy): await background spend counters before asserting by @joshua-berri in #51249
chore(databricks): price claude-fable-5-2 from the proprietary pricing page by @devin-ai-integration[bot] in #51232
feat(azure_ai): support gpt-image edits on Azure AI Foundry by @Bytechoreographer in #49650
fix(proxy): keep a cost settings update from being undone by an in-flight config reload by @devin-ai-integration[bot] in #51273
chore(cost-map): source the fireworks deepseek v4 flash deprecation and name its replacement by @devin-ai-integration[bot] in #51285
test(guardrails): expect the content_filter finish on a blocked chat stream by @devin-ai-integration[bot] in #51281
fix(llms-types): drop the duplicate compact-2026-09-04 arm that fails rust-lint on main by @kerry-berri in #51280
perf(router): list a group's deployments once per registry change on the acompletion path by @narmstrong-berri in #50749
feat(guardrails): add the CollieAi guardrail integration by @collieai-dev in #49224
fix(anthropic_messages): bound the detached pump so it cannot outlive its stream by @nuernber in #51268
fix(auth): allow video retrieval after budget exhaustion by @joshua-berri in #51227
fix(proxy): bill anthropic pass-through calls at the deployment's custom price by @joshua-berri in #51267
ci(unit): split proxy-infra-root in two, isolate the child-python tests, make its file exclusions take effect by @kerry-berri in #51316
fix(gemini): attribute /gemini batchGenerateContent batches to the calling key by @kerry-berri in #51279
ci(unit): deselect test_unit_test_proxy_hooks from the q-z shard it matches, pin the dead proxy-infra-root codecov flag by @kerry-berri in #51324
test(integration): remove the openapi docs gate detached worktree test that cannot run offline by @mateo-berri in #51300
feat(gemini): add gemini-decision-1, the Decisions API model priced at $0.10 per 1M input tokens by @devin-ai-integration[bot] in #51215
feat(proxy): run spend log cleanup from a DB-backed scheduled job table by @jesus-berri in #50131
feat(ui): add Activity tabs to resource details by @moe-berri in #50630
feat(alerts): add persisted alerts and status APIs by @moe-berri in #50658
feat(alerts): detect daily changes in recorded usage by @moe-berri in #50666
feat(ui): show performance alerts in the Alerts page by @moe-berri in #51104
feat(alerts): enable monitoring by default with a five-alert free allowance by @moe-berri in #51123
feat(skills): add admin-only version restore by @tin-berri in #51309
chore: move config and docs off the repo root by @yassin-berriai in #50966
feat(gemini): support gemini-decision-1 on /v1/decisions and /v1/systemone by @mateo-berri in #51220
fix(proxy): reject unknown guardrail names on key, user, team and project writes by @mrinal-berri in #51080
fix(cost-map): drop the duplicate gemini/gemini-decision-1 entry that fails the root unit job on main by @yuneng-berri in #51341
test(providers): expect Rust routing for quiet Messages streams by @devin-ai-integration[bot] in #51346
test(observability): include CollieAi in guardrail scope settings by @devin-ai-integration[bot] in #51345
ci(integration): install the Rust gateway and fail forwarded_proxy when Rust is not in front by @mateo-berri in #51307
chore: move deployment sources under litellm/deploy by @yassin-berriai in #50981
chore: move ui and vscode-extension under litellm and delete docs by @yassin-berriai in #50992
chore: move workspace packages under litellm by @yassin-berriai in #51000
chore: move scripts and runtime sources under litellm by @yassin-berriai in #51006
ci: fail when the repo root grows past 15 entries by @yassin-berriai in #51007
fix(alerts): stop detecting traffic drops by @moe-berri in #51338
feat(rate-limit): team_model_limit_borrows_idle_capacity_from_other_teams lets a team's model TPM and RPM limits borrow idle capacity in the dynamic rate limiter by @kerry-berri in #51328
fix(fal_ai): give MiniMax H3 Max video endpoints their own resolution profile by @kerry-berri in #51348
fix(anthropic): close strict JSON schemas on the Messages to Responses bridge by @kenneth-berri in #51337
test(providers): expect Rust routing for idle Messages streams by @devin-ai-integration[bot] in #51364
feat(rate-limit): team_model_limit_max_borrow_multiple caps how far a team borrows idle model capacity by @kerry-berri in #51359
fix(gemini): add batch and priority image output token prices to gemini/gemini-2.5-flash-image by @devin-ai-integration[bot] in #51368
ci: stop PR runs from thrashing the Actions cache and stop smoke compiling from source by @yuneng-berri in #51343
feat(dynamic_rate_limiter_v3): steer flexible teams to a permitted alternate when their preferred model is saturated by @kerry-berri in #51350
feat(dynamic_rate_limiter_v3): priority_class_share_splits_equally_among_active_teams splits a priority class share among the teams active on a model by @kerry-berri in #51366
fix(rate-limit): hold router fallbacks, nested calls, WebSocket turns and oversized prompts to a borrowing team's model share by @kerry-berri in #51367
feat(model-catalog): read output_cost_per_billable_unit in the rust registry check by @kerry-berri in #51365
chore(cost-map): rolling registry audit 2026-11-12, xAI retired image rates, Kling 4, GMI rows and -latest aliases by @devin-ai-integration[bot] in #51395
feat(skills): let authors edit and withdraw pending submissions by @tin-berri in #51312
chore(cost-map): add 7 fal_ai rows for seedream 5.5, flux virtual try-on, patina material and vidu templates (fifth fal-ai sync) by @devin-ai-integration[bot] in #51407
feat(replicate): add video generation, image generation and text to speech through the predictions API by @kerry-berri in #51360
feat(mcp): cache upstream tool lists with an admin-set lifetime by @shrey-berri in #51160
fix(auth): enforce customer budget rpm/tpm limits under JWT auth by @mateo-berri in #51405
feat(fal_ai): add Nano Banana 2.1, bill it from fal billable units, fix google/ routing by @devin-ai-integration[bot] in #51410
fix(messages): name the deployment on a rust stream's transport error by @narmstrong-berri in #51358
fix(token-counter): count Anthropic server tool history by @kenneth-berri in #51334
feat(ui): add daily model usage to customer pages by @ishaan-j-berri in #51419
feat(proxy): publish opted-in pass-through endpoints on the Model Hub by @devin-ai-integration[bot] in #43905
feat(gemini): add gemini/gemini-4-hydrogen from the Gemini API pricing page by @devin-ai-integration[bot] in #51423
refactor(ui): remove standalone Model Leaderboard by @moe-berri in #51426
feat(alerts): add management controls and opt-in daily summaries by @moe-berri in #51418
fix(skills): honor dashboard user and team grants by @tin-berri in #51416
fix(proxy): fetch blog posts without blocking the event loop by @yassin-berriai in #51369
feat(vertex_ai): add vertex_ai/gemini-4-hydrogen from the Vertex pricing page by @devin-ai-integration[bot] in #51432
test(providers): expect rust for anthropic messages after rust-only cutover by @narmstrong-berri in #51351
fix(spend-logs): mask token-shaped secrets in stored request and resp… by @krrish-berri-2 in #48968
fix(otel): report the provider model in experimental GenAI traces by @joshua-berri in #51430
fix(azure): strip GPT-5 routing marker from Responses deployments by @joshua-berri in #51438
fix(messages): keep signed empty thinking blocks on rust by @narmstrong-berri in #51349
fix(messages): hand rust only the anthropic metadata user_id by @narmstrong-berri in #51355
fix(anthropic_messages): preserve first-round web search spend logs by @joshua-berri in #51441
fix(token-counting): count mid-conversation tool changes by @joshua-berri in #51447
test(integration): point database tests at the moved litellm/proxy-extras package by @devin-ai-integration[bot] in #51453
fix(otel): distinguish missing usage from prompt estimates on chat failures by @joshua-berri in #51443
fix(model_prices): add Azure DeepSeek V4.1 Flash 1017 and two Perplexity Agent API rows by @devin-ai-integration[bot] in #51444
test(messages): expect rust to drop a null thinking signature on retry by @narmstrong-berri in #51461
fix(proxy): refuse blocked and over-limit keys before reading the request body by @kerry-berri in #51428
fix(ui): serve navigation payloads under the ui prefix by @mrinal-berri in #50758
feat(model-catalog): read output_cost_per_video in the rust registry check by @kenneth-berri in #51433
feat(access-groups): grant vector stores through access groups by @shrey-berri in #50290
fix(openai): point gpt-image-1-mini, gpt-image-1.5 and chatgpt-image-latest at gpt-image-2.6-sunburst as successor by @devin-ai-integration[bot] in #51475
fix(ui): remove the Chargebacks page from Customer Portal by @tin-berri in #51474
feat(proxy): customer org, team and member hierarchy with budget templates by @ishaan-j-berri in #51459
feat(fal_ai): bill flat per-video fal.ai models with output_cost_per_video by @kenneth-berri in #51225
feat(cost-map): load LITELLM_MODEL_COST_MAP_URL from gs:// with Google application default credentials by @jesus-berri in #51108
feat(dynamic_rate_limiter_v3): prometheus share, borrowed and throttle metrics per team and model by @kerry-berri in #51417
fix(dynamic_rate_limiter_v3): a borrower gets its full share back once the model saturates by @kerry-berri in #51415
fix(fal_ai): bill Luma Ray v3.2 videos at fal's 540p default resolution by @kerry-berri in #51477
feat(ui): set per-threshold budget alert recipients on virtual keys by @jesus-berri in #51061
fix(embeddings): unwrap extra_body for providers with their own embedding handlers by @kenneth-berri in #51335
fix(bedrock): send max_tokens and priority in Cohere embedding requests by @kenneth-berri in #51435
chore: bump litellm-enterprise 0.1.83 -> 0.1.84, litellm-proxy-extras 0.4.114 -> 0.4.115, litellm 1.110.0 -> 1.111.0 by @yucheng-berri in #51486

New Contributors

@apex-mochen made their first contribution in #43577
@raunakpandey1 made their first contribution in #49806
@tedswinyar made their first contribution in #49183
@kenneth-berri made their first contribution in #50764
@giladk1221 made their first contribution in #50883
@taxfree-python made their first contribution in #46845
@nurlanmnn made their first contribution in #50979
@satwik-bankapur made their first contribution in #49914
@collieai-dev made their first contribution in #49224

Full Changelog: v1.110.0-dev.2...v1.111.0-dev.1