Commit graph

19665 commits

Author SHA1 Message Date
shivam
f0e87f2457 fix(router): keep refusal gates closed once every fallback entry was tried
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-21 19:50:07 +00:00
yuneng
d69eb7f095 fix(proxy): fall back to the credentials table when validating litellm_credential_name
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-21 19:47:27 +00:00
tin-berri
457b01e96d
Merge pull request #42055 from BerriAI/litellm_prompt_caching_request_table
feat(ui): show prompt caching requests and net savings
2026-09-21 12:45:11 -07:00
Mateo Wang
e8d97d381a
Merge pull request #42080 from BerriAI/litellm_prompt_caching_affinity_lookback
fix(router): keep prompt caching affinity when the breakpoint moves
2026-09-21 12:44:34 -07:00
yassin
594d97e7e7 fix(edenai): restore unrelated files the squashed PR commit had reverted
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-21 19:42:37 +00:00
yassin
889b8fb220 Merge remote-tracking branch 'origin/main' into litellm_add_edenai_provider 2026-09-21 19:40:48 +00:00
Joshua Valluru
e208b4e89e fix(mcp): keep explicit legacy sampling callers isolated 2026-09-21 12:40:12 -07:00
Joshua Valluru
68074da1d1 test(mcp): cover stable server ordering and sort priorities 2026-09-21 12:39:47 -07:00
mateo-berri
057c45f23f test(e2e): send no-cache on cacheable SDK calls and accept Bedrock's 400 on the responses leg
The deleted wrapper put cache: {"no-cache": true} on every request body, so the
gateway's response cache never answered a re-sent prompt. The SDKs send nothing
of the sort, and the mid-conversation prompt-cache priming loop re-sends an
identical body until the provider reports a warm cache, which a cached reply
never does. NO_PROXY_CACHE in sdk_clients.py restores the field as extra_body
on every messages, responses, completions, and embeddings call.

The wrapper also returned a 4xx as a value where the SDKs raise. The Bedrock
safety_identifier test judges the captured Converse body, and Claude on Bedrock
rejects the forwarded field with a 400, so the /v1/responses leg now suppresses
openai.BadRequestError the way the chat leg carries the same 400 as a Result.
2026-09-21 12:39:16 -07:00
yucheng
4e388e6aea refactor(proxy): inject scheduled job shutdown timeouts
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-21 19:37:28 +00:00
Yuneng Jiang
86c4021e83
Merge branch 'main' into litellm_config_read_source
The shared settings-store work this branch was stacked on landed on main,
along with source reporting for /config/field/info and /config/list.
Resolved by taking main for the store, the config routes and their tests,
and re-applying only the read surfaces main does not cover:
/alerting/settings, /get/ui_settings and /router/settings.
2026-09-21 12:36:15 -07:00
yucheng
9c411dd6f2 refactor(proxy): make scheduled job shutdown timeouts configurable via env
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-21 19:35:29 +00:00
joshua-berri
211ff96943
Merge pull request #39189 from BerriAI/litellm_mcp_list_pagination_lit5594
fix(mcp): paginate prompt and resource discovery
2026-09-21 19:35:17 +00:00
Mateo Wang
0e7cf5113e
Merge pull request #42069 from BerriAI/litellm_redacted_thinking_prompt_caching_pin
fix(token_counter): count replayed redacted_thinking blocks so prompt_caching keeps pinning
2026-09-21 12:35:14 -07:00
yassin
7dc9609109 Merge branch 'main' into litellm_add_edenai_provider 2026-09-21 19:34:49 +00:00
Joshua Valluru
f5d511b3f9 Merge remote-tracking branch 'origin/main' into litellm_mcp_server_list_stable_order 2026-09-21 12:33:27 -07:00
Joshua Valluru
d2f30a77fd test(mcp): preserve toolset scope across explicit context 2026-09-21 12:31:48 -07:00
Yujong Lee
efcafa7f12 fix(rust): invalidate changed Redis pool settings 2026-09-21 12:30:17 -07:00
mateo-berri
deca6aea79 test(bedrock): assert safeguard_results survive message_start and check the beta by membership 2026-09-21 12:30:11 -07:00
Mateo Wang
d8267d507d
Merge pull request #41956 from BerriAI/litellm_explicit_cache_injection_points_survive_client_marks
fix: apply configured cache_control_injection_points beside client cache_control marks
2026-09-21 12:28:02 -07:00
Joshua Valluru
a835e75620 fix(mcp): handle split UTF-8 routing previews in place 2026-09-21 12:24:19 -07:00
Joshua Valluru
1098604ed6 refactor(mcp): extract explicit operation context and dispatch 2026-09-21 12:24:15 -07:00
yuneng
177021b2ac fix(proxy): gate explicit credential detach only on PATCH /model/{id}/update
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-21 19:22:57 +00:00
yuneng
10d343c3ee fix(proxy): detach stored credential when model editor selects None (LIT-7597)
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-21 19:22:57 +00:00
mateo-berri
0e16050100 fix(anthropic): forward Claude Code safeguards and dangerous-tool-use beta to Bedrock Invoke and Vertex on /v1/messages
Claude Code's server-side auto-mode classifier sends a `safeguards` body field
together with the `dangerous-tool-use-2026-09-03` beta. PR #42152 made the
first-party anthropic route pass them through, but the beta header mapping
left the other two Claude platforms at null, so Bedrock Invoke dropped both
(classifier silently disabled) and Vertex forwarded the body field without
the beta, which the platform rejects with "safeguards: Extra inputs are not
permitted" (a 400 Claude Code hides by retrying without them).

Map the beta for bedrock and vertex_ai in the beta headers config and add
`safeguards` to the Bedrock Invoke request allowlist so the pair reaches
both platforms unchanged. Nothing is injected: a client that sends
`safeguards` without the beta still gets the platform's 400, exactly as
api.anthropic.com answers it.
2026-09-21 12:21:50 -07:00
Mateo Wang
e7e4df9098
Merge pull request #42041 from BerriAI/litellm_azure_ai_gpt5_tools_responses_bridge
fix(azure_ai): bridge gpt-5.4+ function tools with reasoning to the Foundry Responses API
2026-09-21 12:21:02 -07:00
mateo-berri
70be37a73c Merge branch 'devin_ai_fix_azure_cancellederror_35329' of https://github.com/BerriAI/litellm into devin_ai_fix_azure_cancellederror_35329
Some checks failed
LiteLLM Rust / rust-lint (push) Has been cancelled
LiteLLM Rust / rust-test (push) Has been cancelled
LiteLLM Rust / rust-wheel (push) Has been cancelled
Terraform Modules / fmt, validate, test (aws) (push) Has been cancelled
Terraform Modules / fmt, validate, test (gcp) (push) Has been cancelled
Terraform Provider / gofmt, vet, build, test (push) Has been cancelled
Terraform Provider / Provider endpoints vs proxy OpenAPI schema (push) Has been cancelled
# Conflicts:
#	tests/e2e/router/reliability_support.py
#	tests/e2e/router/test_reliability_cancel_on_disconnect_e2e.py
#	tests/test_litellm/llms/azure/test_azure.py
2026-09-21 12:17:20 -07:00
Joshua Valluru
bf5dff8986 chore: sync MCP UTF-8 fix with main 2026-09-21 12:15:57 -07:00
mateo-berri
cf00ab1bf8 fix: rename the mainland China brand to Qianwen AI Platform 2026-09-21 12:13:43 -07:00
shivam
86960fb127 fix(router): walk every entry of a fallback list after a mid-stream failure
A fallback hop that dies before its first chunk surfaces inside the streaming
iterator, where the chain lookup is keyed by the hop's own group. That group has
no chain of its own, so the remaining entries of the original list were never
tried. Resume the original group's chain as the last lookup key; attempted_targets
already skips the entries that were tried.

Resolves LIT-7400

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-21 19:11:38 +00:00
kerry
9b54c4b077 refactor(fal_ai): bill flux dev per 1024x1024 megapixel and drop ImageResponse retyping
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-21 19:10:33 +00:00
kerry
adc4e6a132 fix(fal_ai): price images from the dimensions fal returns
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-21 19:10:32 +00:00
ryan
2074273faf test(proxy): give the faked login user the breach-check columns
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-21 19:06:00 +00:00
kerry
8d73ce756a feat(fal_ai): add MiniMax H3 text-to-video and reference-to-video
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-21 19:03:00 +00:00
Mateo Wang
9a90adad32
Merge pull request #42275 from BerriAI/litellm_claude_platform_messages_beta_passthrough
fix(bedrock): forward anthropic-beta headers verbatim on the Claude platform messages path
2026-09-21 12:01:39 -07:00
Yassin Kortam
5673f67727
Merge pull request #42022 from BerriAI/litellm_redis_durable_spend_log_buffer
fix(proxy): park requeued spend logs in Redis so they survive a pod restart during a DB outage
2026-09-21 14:01:34 -05:00
mateo-berri
2fe5c8990e fix(bedrock_mantle): bill Mantle's un-versioned Claude ids from a Mantle cost row
Mantle serves anthropic.claude-haiku-4-5 without the dated -20251001-v1:0
suffix the Bedrock row carries, so the native route billed it at 0. Add a
bedrock_mantle/anthropic.claude-haiku-4-5 row and let a
bedrock_mantle/<region>/<model> name fall back to the region-free
bedrock_mantle/<model> row before the provider-prefixed lookup. Also
satisfy the mutable-collection gate in the native messages transformation.
2026-09-21 12:00:38 -07:00
mateo-berri
4eed951e6f Merge commit '36b8be7d81b' into litellm_mantle_native_anthropic_messages_b4dc
# Conflicts:
#	tests/test_litellm/llms/anthropic/experimental_pass_through/messages/test_anthropic_experimental_pass_through_messages_handler.py
2026-09-21 12:00:37 -07:00
mateo-berri
b51e80a4a8 Merge remote-tracking branch 'origin/main' into devin_ai_fix_azure_cancellederror_35329 2026-09-21 11:59:56 -07:00
mateo-berri
fda7a078d3 test(e2e): client hang-up under cancel_on_disconnect never benches the Azure deployment 2026-09-21 11:59:55 -07:00
mateo-berri
5ea4fe620f test(router): give each prompt caching check test a fresh callback registry 2026-09-21 11:55:58 -07:00
moe-berri
a83773cfa5
Merge pull request #41886 from BerriAI/litellm_jev_autorouter_launch_1789767495
feat(auto-router): add JEV classifier alongside LLM classifier
2026-09-21 11:55:44 -07:00
mateo-berri
f549c87091 Merge branch 'main' into litellm_prompt_caching_affinity_lookback 2026-09-21 11:55:40 -07:00
Mateo Wang
7cb884cd31
Merge pull request #42120 from BerriAI/litellm_cadence_58065d4_google_genai_proxy_master_key
test(google): boot the unified Google proxy fixture with a real master key
2026-09-21 11:54:03 -07:00
ryan
6bfbc7dba4 test(auth): pass throttle to authenticate_user in reset-required tests
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-21 18:53:39 +00:00
ryan
3fe405a5cc feat(auth): breached password detection, self-service change-password and forced password reset
Cherry-pick of merge commit b3882d8e43 (PRs #39321, #39562, #40107), which landed on litellm_internal_staging instead of main.

Co-authored-by: ojensen-berri <ojensen@berri.ai>
Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-21 18:48:35 +00:00
kerry-berri
12379aa1e3
Merge pull request #42095 from BerriAI/litellm_fal_gpt_image_25_flux_dev_edits
feat(fal_ai): add gpt-image-2.5 flare/sunburst, flux/dev and image edits
2026-09-21 11:45:45 -07:00
Devin AI
a3f1956090 test(e2e): client disconnect must not bench the Azure deployment it cancelled
Live proxy with cancel_on_disconnect, two-deployment group, generic allowed_fails=0, red at the pre-fix handler and green with the bare raise

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-09-21 18:39:37 +00:00
mateo-berri
e51ccbc759 fix(bedrock): forward anthropic-beta headers verbatim on the Claude platform messages path 2026-09-21 11:39:26 -07:00
mateo-berri
f567fe230e fix: reserve cap slots for direct marks on /v1/messages when extra_body unmarks them 2026-09-21 11:39:25 -07:00