Skip to content

Releases

v1.108.10

Choose a tag to compare

@litellm-release litellm-release released this 16 Nov 21:57
5d6a139

Verify Docker Image Signature

cosign verify \
  --certificate-identity https://github.com/BerriAI/project-releaser/.github/workflows/release.yml@refs/heads/main \
  --certificate-oidc-issuer https://token.actions.githubusercontent.com \
  ghcr.io/berriai/litellm:v1.108.10

What's Changed

New Contributors

Full Changelog: v1.108.9...v1.108.10

v1.111.0-dev.4

v1.111.0-dev.4 Pre-release
Pre-release

Choose a tag to compare

@litellm-release litellm-release released this 15 Nov 04:17
61483d8

Important

Signing has moved to keyless cosign. Since v1.111.0-dev.3, releases are signed with Sigstore keyless signing (GitHub OIDC) instead of the cosign.pub key.
cosign verify --key ... does not work for this release. Use the command below. Earlier releases still verify with the key.

Verify this release

cosign verify \
  --certificate-identity https://github.com/BerriAI/project-releaser/.github/workflows/release.yml@refs/heads/main \
  --certificate-oidc-issuer https://token.actions.githubusercontent.com \
  ghcr.io/berriai/litellm:v1.111.0-dev.4

Same command for docker.io/litellm/litellm and the Helm chart at ghcr.io/berriai/litellm/chart/litellm:1.111.0-dev.4

What's new in 1.111.0-dev.4

138 PRs. 18 added, 69 fixed, 11 changed, 3 improved, 1 removed. 2 need action

Action required

  • Logging and observability: Set general_settings.redact_tokens_in_spend_logs: true to keep token scanning in spend logs (#51729)
  • AWS Bedrock: Remove unsupported service_tier values from Bedrock requests (#50516)

Highlights

  • Admin UI: Added customer organization, team and member management in Customer Management (#51521, #51661, #51713)
  • Vector stores: Added OpenSearch vector store filters support and hybrid keyword plus vector search (#51131)
  • Keys, teams and auth: Added budget_limits windows on internal users (#51763)
  • Keys, teams and auth: Added email warnings to key owners before opted-in virtual keys expire (#51482)
  • SDK: Fixed hang and RecursionError on every completion call when litellm.model_fallbacks is set (#51817)
  • Deployment: Fixed proxy boot failure on PostgreSQL 12 without pgcrypto, broken since v1.110.0-rc.1 (#51868)
  • Performance: Improved streaming latency by closing streams right after the last event (#51487)
  • Responses API: Fixed Responses fallback replaying tool calls that were already delivered (#51683)
  • Azure: Fixed 400 on azure/ Responses and chat calls to Foundry project /openai/v1 bases (#51767)

All changes

OpenAI (2)
  • Fixed batch prices for gpt-image-2.6-sunburst and gpt-image-2.6-flare, billed at twice the rate before (#51647)
  • Fixed 500 and unbilled first round on web search with openai/responses/ chat completions (#51802)
Azure (5)
  • Fixed retirement dates for azure_ai/claude-haiku-4-5 and azure_ai/claude-opus-4-5 to 2027-03-16 (#51716)
  • Fixed reasoning_effort 400 and dropped reasoning_content for Kimi K3 on azure_ai (#51747)
  • Fixed server tracebacks reaching clients and alerts when an Azure stream is cut (#51760)
  • Fixed 400 on azure/ Responses and chat calls to Foundry project /openai/v1 bases (#51767)
  • Changed /v1/messages for Claude on Azure AI Foundry to run on the Rust gateway (#51721)
AWS Bedrock (5)
  • Fixed 500 on Bedrock Converse requests with s3:// image, document or video URLs (#50577)
  • Fixed aborted Bedrock Invoke /v1/messages streams still reading from Bedrock (#50657)
  • Fixed Bedrock vector store search 400 on OpenAI-shape filters and ignored drop_params (#51291)
  • Fixed Titan embedding calls on Bedrock /invoke passthrough not being billed or logged (#51701)
  • Changed Bedrock to return 400 for unsupported service_tier values and to forward reserved (#50516)
Vertex AI and Gemini (9)
  • Added key, team, org and project labels on Vertex AI requests (#50732)
  • Fixed wrong default max_tokens for versioned Claude ids on Vertex AI (#40376)
  • Fixed shutdown dates on Gemini TTS and Live preview models to match Google's deprecations page (#51657)
  • Fixed PDF embeddings on gemini-embedding-2 reporting 0 tokens and $0 cost (#51673)
  • Fixed document_ocr, audio_track_extraction and auto_truncate being dropped on Gemini embeddings (#51686)
  • Fixed streamed Gemini TTS on chat completions returning no audio in delta.audio (#51750)
  • Fixed Vertex 400 for tools with a top-level anyOf, oneOf or allOf schema (#51835)
  • Improved Vertex RAG Engine vector store searches by reusing one Google token (#51239)
  • Removed Flex prices from three Vertex AI Gemini 4 rows (#51735)
Other providers (13)
  • Added Vercel AI Gateway models for Decisions requests with cost tracking (#50642)
  • Added GitHub App authentication with PEM upload for GitHub Copilot (#51427)
  • Added fal.ai Pixverse video billing by clip duration and resolution, billed at $0 before (#51679)
  • Added github_copilot/gemini-4-hydrogen to the model cost map (#51696)
  • Added the 4K output price for runwayml/seedance2_5, billed at the 720p rate before (#51697)
  • Fixed OpenRouter streams emitting an empty choice after finish_reason: stop (#50210)
  • Fixed Cohere embeddings ignoring COHERE_API_BASE (#50249)
  • Fixed Hugging Face chat ignoring HUGGINGFACE_API_BASE and crashing with env endpoints and huggingface/tgi (#51449)
  • Fixed low and medium reasoning effort being sent as high for mistral/mistral-large-4 (#51693)
  • Fixed 500 when streaming opencode_go Messages-protocol models through chat completions (#51715)
  • Fixed together_ai/* listing retired Together models that answer 400 (#51827)
  • Changed OpenRouter failed non-streaming completions to raise instead of returning success (#50880)
  • Changed Mistral Magistral to receive native reasoning levels instead of prompt instructions (#51329)
Chat completions (1)
  • Fixed 500 on gateway chat completions when Claude returns a thinking block (#51774)
Responses API (2)
  • Fixed Responses fallback replaying tool calls that were already delivered (#51683)
  • Fixed 500 when allowed_openai_params is a string or tuple on Responses models, broken since v1.109.0-rc.1 (#51751)
Messages API (2)
  • Fixed synchronous iteration of streamed /v1/messages calls routed to OpenAI or Azure (#51305)
  • Fixed gateway ignoring a deployment's thinking, system and tools params on Messages requests (#51834)
Batches and files (2)
  • Added per-model enqueued token limits for batches on projects (#51608)
  • Fixed 404 retrieving or deleting a managed file when the request also names a model (#51780)
Vector stores (1)
  • Added OpenSearch vector store filters support and hybrid keyword plus vector search (#51131)
Token counting (4)
  • Added native upstream token counting for Mistral (#50653)
  • Fixed token counts ignoring thinking, reasoning effort and tool_choice (#51079)
  • Fixed local token counts ignoring the request's tool_choice on the count routes (#51133)
  • Changed /v1/messages/count_tokens to answer 400 instead of 500 for messages missing role or content (#51284)
MCP (2)
  • Fixed MCP gateway sign-in ignoring the servers selected on the connect page (#50973)
  • Fixed resources/read failing on the aggregate /mcp endpoint with two or more servers (#51462)
Guardrails (1)
  • Changed Straiker blocked prompts to return 200 with the block message instead of 400 (#51401)
Caching (2)
  • Fixed in-memory cache keeping expired keys and panicking on very long TTLs (#51388)
  • Fixed cache_control being dropped on tool plus reasoning chat requests sent over Responses (#51783)
Routing and fallbacks (4)
  • Fixed Complexity Router dropping temperature and output-token limits from routed prompts (#51702)
  • Fixed /model_group/info missing key aliases and team model aliases (#51754)
  • Fixed wildcard aliases over unknown orgs listing every catalog model id (#51805)
  • Changed over-tpm requests to fail at once with an error naming the request size (#51838)
Cost tracking and pricing (3)
  • Added cost map rows for Soniox v6 and four DeepInfra models, and corrected retirement dates (#51620)
  • Added fal_ai Seedream v5 Pro US text-to-image and edit pricing (#51881)
  • Fixed cost maps with a byte order mark failing to load and register_model(url) calling out in offline mode (#51545)
Logging and observability (8)
  • Added a proxy-level default langfuse_span_scope for key and team Langfuse destinations (#51726)
  • Fixed Langfuse environment, release and version missing on langfuse_otel traces with OTel v2 (#51385)
  • Fixed system prompt blocks sent to /v1/messages missing from logged messages (#51699)
  • Fixed tool and auto-router usage being lost after database disconnects (#51706)
  • Fixed missing litellm_dynamic_limiter_share_tokens gauge for uncapped teams served while the model is idle (#51709)
  • Fixed spend logs dropping caller request metadata on auth failures (#51745)
  • Fixed duplicate async success logs for sync streamed openai/responses/ chat calls (#51787)
  • Changed spend log token redaction to be off by default (#51729)
Keys, teams and auth (18)
  • Added email warnings to key owners before opted-in virtual keys expire (#51482)
  • Added budget_limits windows on internal users (#51763)
  • Fixed carried end-user spend being lost when a budget reset is retried (#51143)
  • Fixed 401 Expired Key in Claude Code after a lite login --pkce --config-claude key expires, broken since v1.102.0 (#51319)
  • Fixed customer org provisioning retries failing and duplicate org aliases breaking requests (#51607)
  • Fixed customer budget alerts not firing for members and ignoring team budget ...
Read more

v1.111.0-dev.3

v1.111.0-dev.3 Pre-release
Pre-release

Choose a tag to compare

@yuneng-berri yuneng-berri released this 15 Nov 05:09
61483d8

Warning

This release was only partly published. Use v1.111.0-dev.4 instead.
It is built from the same commit and fixes the publishing problems below.

Important

Signing has moved to keyless cosign. This is the first release signed with Sigstore keyless signing (GitHub OIDC) instead of the cosign.pub key.
cosign verify --key ... does not work for this release or later ones. Use the command below. Earlier releases still verify with the key.

Known issues, fixed in v1.111.0-dev.4

  • pip install "litellm[proxy]==1.111.0.dev3" failed at first, because the litellm-proxy-extras and litellm-enterprise versions it pins were not on PyPI yet. They are published now, so the install works
  • The enterprise package repository is missing the litellm sdist and the litellm-proxy wheels for this version

The Docker image, Helm chart and PyPI wheels of this release are otherwise the same code as v1.111.0-dev.4. See its notes for what changed

Verify this release

cosign verify \
  --certificate-identity https://github.com/BerriAI/project-releaser/.github/workflows/release.yml@refs/heads/main \
  --certificate-oidc-issuer https://token.actions.githubusercontent.com \
  ghcr.io/berriai/litellm:v1.111.0-dev.3

v1.110.0-rc.3

v1.110.0-rc.3 Pre-release
Pre-release

Choose a tag to compare

@litellm-release litellm-release released this 15 Nov 01:10
3510cb6

Verify this release

cosign verify \
  --key https://raw.githubusercontent.com/BerriAI/litellm/e82a3c646908ba0a6ee0beddc4310d7a6cac4065/cosign.pub \
  ghcr.io/berriai/litellm:v1.110.0-rc.3

Same command for litellm-database, litellm-non_root and the docker.io/litellm/* images

What's new in 1.110.0-rc.3

7 PRs. 6 fixed, 1 changed. No action needed to upgrade

Changes

  • Other providers: Fixed Hugging Face chat sending requests to None or crashing when the endpoint comes from HUGGINGFACE_API_BASE (#51449)
  • Other providers: Fixed unmapped Fireworks AI models sized like 1p5b or 0p6b billing at double the up-to-4B rate (#51451)
  • Other providers: Fixed 500 when streaming OpenCode Go Messages-protocol models through chat completions (#51715)
  • Responses API: Fixed Responses WebSocket requests dropping the team's model alias and landing on another deployment (#51380)
  • Token counting: Fixed follow-up requests with Anthropic server tool history, such as web search, skipping the context-window check (#51334)
  • Routing and fallbacks: Fixed 500 when a request sends a string fallback override such as "fallbacks": "default", broken since v1.110.0-dev.1 (#51502)
  • Deployment: Changed opentelemetry, dnspython, tornado and soupsieve to the newer versions that main already ships (#51854)

Backport #51514 Β· Backport #51854 Β· Backport #51858 Β· Full changelog

1.108.9 (LTS)

Choose a tag to compare

@litellm-release litellm-release released this 14 Nov 22:00
82ae95f

Tip

The lts image tag always points to the latest LTS release. Pull ghcr.io/berriai/litellm:lts to stay on the newest LTS patch

Verify this release

cosign verify \
  --key https://raw.githubusercontent.com/BerriAI/litellm/e82a3c646908ba0a6ee0beddc4310d7a6cac4065/cosign.pub \
  ghcr.io/berriai/litellm:v1.108.9

Same command for litellm-database, litellm-non_root and the docker.io/litellm/* images

What's new in 1.108.9 (LTS)

10 PRs. 2 added, 7 fixed, 1 changed. No action needed to upgrade

Changes

  • Azure: Fixed Kimi K3 on Azure AI Foundry losing its earlier reasoning after tool calls and refusing reasoning_effort (#51747)
  • Azure: Fixed 400 on Foundry project /openai/v1 bases and 404 on resource-level /openai/v1 Responses bases (#51767)
  • AWS Bedrock: Fixed 500 on s3:// image, PDF and video URLs on Converse models, which now go to Bedrock as S3 sources (#50577)
  • AWS Bedrock: Fixed AWS 400 on Bedrock vector store search with OpenAI-style filters (#51291)
  • Vertex AI and Gemini: Added vertex_request_label_fields to label Vertex calls with the caller's key, team, org and project (#50732)
  • Vertex AI and Gemini: Fixed 400 Extra inputs are not permitted when sending labels to Claude on Vertex AI (#48753)
  • Other providers: Added Vercel AI Gateway models on /v1/decisions, billed at the cost Vercel reports (#50642)
  • Responses API: Fixed streamed /v1/responses dropping the first answer fragment when it arrives together with reasoning (#51356)
  • Responses API: Fixed cache_control marks being dropped on chat requests with function tools bridged to the Responses API (#51783)
  • Deployment: Changed the Docker images to a newer Wolfi base and updated opentelemetry, dnspython, tornado and soupsieve (#51853)

Backport #50401 Β· Backport #51250 Β· Backport #51292 Β· Backport #51748 Β· Backport #51769 Β· Backport #51786 Β· Backport #51796 Β· Full changelog

v1.107.0-rc.4

v1.107.0-rc.4 Pre-release
Pre-release

Choose a tag to compare

@litellm-release litellm-release released this 14 Nov 23:46
a1a541b

Verify this release

This release candidate is built from a1a541b, the reviewed merge of #51771 into rc/1.107.0. It is published as 1.107.0rc4 on PyPI and v1.107.0-rc.4 on Docker

After the Oct 21 incident (#47826) we replaced the cosign key that signs our Docker images. These images are signed with the new key from commit e82a3c6

cosign verify \
  --key https://raw.githubusercontent.com/BerriAI/litellm/e82a3c646908ba0a6ee0beddc4310d7a6cac4065/cosign.pub \
  ghcr.io/berriai/litellm:v1.107.0-rc.4

Why a new 1.107 release candidate

Some users are still running the 1.107 release candidates and needed to link their GitHub Copilot per-user credentials without access to the Admin UI. This release brings the lite credentials connect command to that line. If you are not on a 1.107 release candidate, 1.109.1 is the current stable

What's new in 1.107.0-rc.4

1 PR. 1 added, 1 changed. No action needed to upgrade

Changes

  • Other providers: Added lite credentials connect to link a per-user GitHub Copilot credential from the terminal without the Admin UI (#50863)
  • Deployment: Changed aiohttp, ddtrace, pypdf and other locked dependencies in the Docker images to newer releases (#51771)

Backport #51771 Β· Full changelog

v1.111.0-dev.2

v1.111.0-dev.2 Pre-release
Pre-release

Choose a tag to compare

@litellm-release litellm-release released this 13 Nov 19:05
ba1257b

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.2

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.2/cosign.pub \
  ghcr.io/berriai/litellm:v1.111.0-dev.2

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

  • chore(cost-map): point the fireworks deepseek v4 flash successor at deepseek-v4p1-flash by @kerry-berri in #51478
  • fix(cost): preserve Bedrock cross-region pricing for Messages by @joshua-berri in #51479
  • test(image): point admin MCP schema push at /app/litellm/schema.prisma by @maciej-berri in #51446
  • feat(proxy): add customer usage tracking and drilldowns by @ishaan-j-berri in #51495
  • fix(dynamic_rate_limiter_v3): a capped borrower keeps its cap on its own counter and gets a fresh share at saturation by @kerry-berri in #51494
  • perf(rust): stop re-serializing the Messages request body in Python logging callbacks by @maciej-berri in #51440
  • feat(ui): add deployment-wide Aria visibility setting by @tin-berri in #51506
  • test(ui): give the skills panel's first render 5s to show the catalog by @kerry-berri in #51509
  • fix(router): validate request fields and ignore scalar fallback overrides by @narmstrong-berri in #51502
  • feat(ui): show customer organization usage and budget utilization by @tin-berri in #51483
  • feat(ui): manage customer organizations, teams, and members by @ishaan-j-berri in #51507
  • test: drop the unit tests that flaked on main this week by @kerry-berri in #51488
  • refactor(proxy): name customer member budget columns budget_id and budget_override_id by @kerry-berri in #51490
  • test(security): finish request accounting before changing Vault config by @devin-ai-integration[bot] in #51515
  • fix(ui): align customer member budget form with API contract by @tin-berri in #51523
  • chore(cost-map): add together_ai/thinkingmachines/Inkling-2-Small from the Together models API by @devin-ai-integration[bot] in #51501
  • chore(cost-map): add fal_ai/google/gemini-omni-flash/v1.2/upscale priced per second of upscaled video by @devin-ai-integration[bot] in #51500
  • fix(dynamic_rate_limiter_v3): read a model's saturation from Redis so every worker agrees on it by @kerry-berri in #51519
  • test(providers): compare sagemaker embedding bodies exactly and assert caller encoding wins by @devin-ai-integration[bot] in #51536
  • refactor(router): move Redis key construction out of router-core into the stores by @yujonglee-berri in #50356
  • feat(redis): serve the health checks' deployment states from a native store by @yujonglee-berri in #50357
  • refactor(router): count the whole router_usage family on the native store by @yujonglee-berri in #50862
  • feat(redis): serve the router's session pins from a native store by @yujonglee-berri in #50870
  • refactor(redis): replace the Python RedisBatch with Rust by @yujonglee-berri in #51383
  • fix(proxy): keep call id, model and key logging on keys refused before the body is read by @mateo-berri in #51525
  • feat(budgets): support annual calendar reset periods by @tin-berri in #51508
  • chore(cost-map): add fal_ai/decart/lucy-5b/image-to-video at $0.15 per video from its fal.ai page by @devin-ai-integration[bot] in #51551
  • feat(proxy): enforce customer org, team and member budgets and limits on the request path by @kerry-berri in #51489
  • test(observability): identify team datastore spans by team id in otel excluded services matrix by @devin-ai-integration[bot] in #51554
  • fix(langfuse): reserve span capacity for failure usage attributes by @joshua-berri in #51538
  • feat(proxy): register unknown customer members in the aliased organization's default team by @ishaan-j-berri in #51546
  • feat(nace): add Drex Max decisions and inline images by @saruar-dev in #48426
  • fix(files): match every spelling of a provider file id in the ownership check by @kerry-berri in #51498
  • feat(proxy): router_settings.fallback_on_saturation walks key and team fallbacks before a model saturates, replacing team model_preferences by @kerry-berri in #51439
  • fix(openai): point gpt-5, gpt-5-mini, gpt-5-nano, gpt-5-pro, o3 and o3-pro at the GPT-6 models as successor by @devin-ai-integration[bot] in #51492
  • fix(openai): record the successor for the gpt-4o-mini audio and realtime preview snapshots by @devin-ai-integration[bot] in #51563
  • fix(utils): register ollama deployment prices from the built-in catalog without calling the provider by @devin-ai-integration[bot] in #51561
  • fix(proxy): import soundfile lazily in the transcribe logging handler by @joshua-berri in #51480
  • fix(router): enforce access groups for pinned deployments by @joshua-berri in #51442
  • fix(interactions): route Vertex lifecycle calls through the selected deployment by @joshua-berri in #51511
  • fix(interactions): bind Deep Research agents and record their costs by @joshua-berri in #51517
  • fix(databricks): record the November 19 retirement and successor for databricks-gemini-3-6-flash by @devin-ai-integration[bot] in #51593
  • fix(deps): bump soupsieve past GHSA-4j63-6c4x-wfg4 and GHSA-mmhr-3c5m-f64f by @devin-ai-integration[bot] in #51585
  • fix(proxy): preserve team aliases in Responses WebSockets by @brianhong-dev in #51380
  • fix(cost-map): mark together_ai/elorian-labs/glimpse-1.0 as supporting function calling and structured outputs by @devin-ai-integration[bot] in #51612
  • fix(proxy): enforce settled customer budgets across workers by @ishaan-j-berri in #51568
  • fix(mcp): treat the mcp_all scope as a wildcard any grant covers by @shrey-berri in #51578
  • feat(access-groups): name group vector stores and list them on team info by @shrey-berri in #51484
  • feat(proxy): customer budget rollup, reset on the budget row's window, named templates, team overrides, multi-window limits and pause_allow_increase by @ishaan-j-berri in #51582
  • feat(proxy): run requests with unknown customer ids untracked instead of refusing them by @ishaan-j-berri in #51548
  • feat(proxy): pick the customer organization's default budget template by name on create and by id on patch by @ishaan-j-berri in #51636
  • feat(proxy): register an unknown customer organization alias as a new organization charged for the request by @ishaan-j-berri in #51637
  • fix(proxy): alert on customer org, customer team, and customer member budgets by @ishaan-j-berri in #51579
  • feat(ui): show and pick vector stores for access groups by @shrey-berri in #51581
  • fix(proxy): display customer identity in spend logs by @ishaan-j-berri in #51596
  • fix(fireworks_ai): read fractional sizes like 1p5b in the size bucket fallback by @Abelo9996 in #51451
  • chore: bump litellm-enterprise 0.1.84 -> 0.1.85, litellm-proxy-extras 0.4.115 -> 0.4.116 by @yucheng-berri in #51643

New Contributors

Read more

1.108.8 (LTS)

Choose a tag to compare

@litellm-release litellm-release released this 13 Nov 20:21
e1476b2

Tip

The lts image tag always points to the latest LTS release. Pull ghcr.io/berriai/litellm:lts to stay on the newest LTS patch

Verify this release

cosign verify \
  --key https://raw.githubusercontent.com/BerriAI/litellm/e82a3c646908ba0a6ee0beddc4310d7a6cac4065/cosign.pub \
  ghcr.io/berriai/litellm:v1.108.8

Same command for litellm-database, litellm-non_root and the docker.io/litellm/* images

What's new in 1.108.8 (LTS)

4 PRs. 3 fixed. No action needed to upgrade

Changes

  • Azure: Fixed 400 on /v1/decisions for azure/ deployments (#51460)
  • Batches and files: Fixed file ownership checks missing other spellings of a Gemini file id, such as files/<id> or its URL (#51503)
  • MCP: Fixed sign-in loop ending in Authentication incomplete on servers that answer mcp_all with detailed scopes, broken since v1.107.0 (#51578)

Backport #51640 Β· Full changelog

v1.106.10

Choose a tag to compare

@litellm-release litellm-release released this 14 Nov 00:34
87aa6f7

Verify this release

cosign verify \
  --key https://raw.githubusercontent.com/BerriAI/litellm/e82a3c646908ba0a6ee0beddc4310d7a6cac4065/cosign.pub \
  ghcr.io/berriai/litellm:v1.106.10

Same command for litellm-database, litellm-non_root and the docker.io/litellm/* images

What's new in 1.106.10

2 PRs. 1 fixed. No action needed to upgrade

Changes

  • MCP: Fixed Google MCP servers asking users to sign in again when the access token expires (#41197)

Backport #51731 Β· Full changelog

v1.111.0-dev.1

v1.111.0-dev.1 Pre-release
Pre-release

Choose a tag to compare

@litellm-release litellm-release released this 13 Nov 03:42
6cb6456

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-...
Read more