scriptoriumadmin
  • Joined on 2026-03-01
scriptoriumadmin synced commits to refs/pull/48472/head at scriptoriumadmin/litellm from mirror 2026-10-30 06:48:58 +00:00
dcf0f62400 fix(guardrails): keep naming pipeline steps that ran at the deployment level, in a later pipeline, or on another request
c984395879 fix(guardrails): drop the steps of later response pipelines a block never reached
b55f03740c fix(guardrails): keep every response step listed on generateContent
05d2a0c63f fix(guardrails): name every response step up front and drop the ones that did not run while nothing has read them
Compare 4 commits »
scriptoriumadmin synced commits to refs/pull/48472/merge at scriptoriumadmin/litellm from mirror 2026-10-30 06:48:58 +00:00
dcf0f62400 fix(guardrails): keep naming pipeline steps that ran at the deployment level, in a later pipeline, or on another request
ccc56d27ad fix(azure): add batch prices to the unversioned o3, o4-mini, o3-mini, gpt-4o and gpt-4o-mini rows (#49182)
b91fc92b6d fix(router): cap the default OTPM reservation and reconcile token counters after each call (#49054)
342ce6ae4a feat(zai): support response_format and max_completion_tokens, add glm-5v-turbo pricing (#48335)
Compare 182 commits »
scriptoriumadmin synced commits to refs/pull/48489/head at scriptoriumadmin/litellm from mirror 2026-10-30 06:48:58 +00:00
f6ecc40326 fix(proxy): clear provider and mapping fields the SSO form drops, keep env out of the row
scriptoriumadmin synced commits to refs/pull/48409/merge at scriptoriumadmin/litellm from mirror 2026-10-30 06:48:57 +00:00
e6a07b249a fix(responses): preserve client tool search during upstream sync
d037d86a94 fix(token_counter)!: keep a loaded Hugging Face tokenizer when a slower download fails (#48916)
c87a25a357 feat(proxy)!: general_settings.claude_code_marketplace_auth_required puts the Claude Code marketplace behind a LiteLLM key (#48980)
e8f0735d8a feat(proxy): let a Claude Code gateway user choose the session team (#49001)
Compare 185 commits »
scriptoriumadmin synced commits to refs/pull/48466/merge at scriptoriumadmin/litellm from mirror 2026-10-30 06:48:57 +00:00
23595f7fe5 fix(bedrock): backfill published regional service tier prices (#49113)
85af2f65c9 refactor(helm)!: connect both charts to a separately deployed Lens (#49169)
44cd858697 fix(cost): bill the requested model's rates when the response names an unpriced model (#48900)
a7030a5a2c fix(vertex_ai): batch Claude models in Anthropic's native row shape (#45715)
Compare 130 commits »
scriptoriumadmin synced commits to refs/pull/48440/merge at scriptoriumadmin/litellm from mirror 2026-10-30 06:48:57 +00:00
4ce724e3df fix(caching): key cached async clients by a per-loop token instead of id(loop) (#48877)
d884167bf2 fix(responses): only merge repeated Codex tools when drop_params is on (#49112)
62b8ee78de ci: run the integration groups on GitHub Actions (#48985)
ffcc4f7a70 fix(otel)!: print OTel v2 spans to stdout only when the operator asks (#48723)
Compare 185 commits »
scriptoriumadmin synced commits to refs/pull/48439/merge at scriptoriumadmin/litellm from mirror 2026-10-30 06:48:57 +00:00
d8f162ea22 fix(guardrails): allow MCP call modes on the generic guardrail API (#48060)
1a2b443a32 ci: run the extensions integration group with 48 workers (#49134)
8d441ebc15 test(otel): give the Langfuse destination tests the otel callback's stdout exporter (#49129)
4ce724e3df fix(caching): key cached async clients by a per-loop token instead of id(loop) (#48877)
Compare 188 commits »
scriptoriumadmin synced commits to refs/pull/48411/merge at scriptoriumadmin/litellm from mirror 2026-10-30 06:48:57 +00:00
b91fc92b6d fix(router): cap the default OTPM reservation and reconcile token counters after each call (#49054)
342ce6ae4a feat(zai): support response_format and max_completion_tokens, add glm-5v-turbo pricing (#48335)
733cfeba93 fix(cost): bill a provider reported usage.cost on every path unless the deployment is priced (#48953)
369f3cb5f0 feat(cost-map): add baseten joey-1 decision model (#49123)
Compare 326 commits »
scriptoriumadmin synced commits to refs/pull/48412/merge at scriptoriumadmin/litellm from mirror 2026-10-30 06:48:57 +00:00
c213234783 feat(vertex_ai)!: use HTTP/2 httpx client for search_api vector store (#40631)
ead221995e refactor(router): move router callers onto the entry points and invert the proxy imports (#48847)
57f6e1b39f fix(proxy): keep the litellm command light so spawned workers answer uvicorn's first health ping (#45039)
3057c47d6d fix(caching): keep reasoning, citations, and audio on streamed cache hits (#32096)
Compare 175 commits »
scriptoriumadmin synced commits to refs/pull/48435/merge at scriptoriumadmin/litellm from mirror 2026-10-30 06:48:57 +00:00
8d441ebc15 test(otel): give the Langfuse destination tests the otel callback's stdout exporter (#49129)
4ce724e3df fix(caching): key cached async clients by a per-loop token instead of id(loop) (#48877)
d884167bf2 fix(responses): only merge repeated Codex tools when drop_params is on (#49112)
62b8ee78de ci: run the integration groups on GitHub Actions (#48985)
Compare 154 commits »
scriptoriumadmin synced commits to refs/pull/48432/merge at scriptoriumadmin/litellm from mirror 2026-10-30 06:48:57 +00:00
c98641fd5a feat(dashscope)!: derive protocol bases from one domain (#47812)
06cdee91da fix(guardrails): scan and rewrite streamed /v1/completions under post_call guardrails (#49035)
961fabbb76 feat(secret_managers): resolve oidc/google api_key references per request (#47703)
6cdb9959df feat(router): opt in to non-chat deployment filtering (#46257)
Compare 70 commits »
scriptoriumadmin synced commits to refs/pull/48420/merge at scriptoriumadmin/litellm from mirror 2026-10-30 06:48:57 +00:00
bb0768f54f refactor(proxy): drop the tag spend note from the spend flush
b7058fb1df test(proxy): cover the shutdown tag spend flush end to end and without doubles
c92767d385 fix(proxy): flush the daily tag spend queue on shutdown
dff474e450 fix(responses): replay the original Responses item order, phase, and boundaries on the next chat turn (#48912)
Compare 77 commits »
scriptoriumadmin synced commits to refs/pull/48420/head at scriptoriumadmin/litellm from mirror 2026-10-30 06:48:57 +00:00
bb0768f54f refactor(proxy): drop the tag spend note from the spend flush
b7058fb1df test(proxy): cover the shutdown tag spend flush end to end and without doubles
c92767d385 fix(proxy): flush the daily tag spend queue on shutdown
dff474e450 fix(responses): replay the original Responses item order, phase, and boundaries on the next chat turn (#48912)
3f01341f0a fix(health)!: openai realtime health check reports healthy with a missing or invalid api key (#41233)
Compare 316 commits »
scriptoriumadmin synced commits to refs/pull/48406/merge at scriptoriumadmin/litellm from mirror 2026-10-30 06:48:56 +00:00
4212f1af27 fix(passthrough): forward OpenAI-Beta and default agents=v1 on the OpenAI passthrough (#49009)
ebf84f8649 fix(caching): read the cache key's model group from the router's metadata for the call (#48917)
41602e318d fix(proxy): stream a guardrail block on /v1/messages as Anthropic events (#49010)
444423a5ca fix(cost): bill a deployment's per-second rates on native Anthropic /v1/messages (#42417)
Compare 148 commits »
scriptoriumadmin synced commits to refs/pull/48409/head at scriptoriumadmin/litellm from mirror 2026-10-30 06:48:56 +00:00
e6a07b249a fix(responses): preserve client tool search during upstream sync
d037d86a94 fix(token_counter)!: keep a loaded Hugging Face tokenizer when a slower download fails (#48916)
c87a25a357 feat(proxy)!: general_settings.claude_code_marketplace_auth_required puts the Claude Code marketplace behind a LiteLLM key (#48980)
e8f0735d8a feat(proxy): let a Claude Code gateway user choose the session team (#49001)
c489aa7117 feat(proxy)!: add suppress_startup_output to turn off startup prints (#47708)
Compare 327 commits »
scriptoriumadmin synced commits to refs/pull/48405/merge at scriptoriumadmin/litellm from mirror 2026-10-30 06:48:56 +00:00
53ce41625f chore(databricks): retirement dates for DeepSeek V4 Flash (0731), V4 Pro (0813) and Inkling (#49124)
d25cde4792 chore(cost-map): add fireworks ultramarine 1 flash us and date deepseek v4 flash (#49146)
326f17594e fix(batches): return real batch token counts next to OpenAI's batch usage fields (#47740)
594f862bf3 fix(streaming): rebuild each choice of an n>1 chat stream on its own (#49002)
Compare 113 commits »
scriptoriumadmin synced commits to refs/pull/48397/head at scriptoriumadmin/litellm from mirror 2026-10-30 06:48:56 +00:00
fe5be2a2c7 fix(responses): reconcile continuation validation with upstream tool hoisting
d884167bf2 fix(responses): only merge repeated Codex tools when drop_params is on (#49112)
62b8ee78de ci: run the integration groups on GitHub Actions (#48985)
ffcc4f7a70 fix(otel)!: print OTel v2 spans to stdout only when the operator asks (#48723)
a54204b583 feat(guardrails): scan sampled stream windows in the background (#48952)
Compare 372 commits »
scriptoriumadmin synced commits to refs/pull/48397/merge at scriptoriumadmin/litellm from mirror 2026-10-30 06:48:56 +00:00
fe5be2a2c7 fix(responses): reconcile continuation validation with upstream tool hoisting
4ce724e3df fix(caching): key cached async clients by a per-loop token instead of id(loop) (#48877)
d884167bf2 fix(responses): only merge repeated Codex tools when drop_params is on (#49112)
62b8ee78de ci: run the integration groups on GitHub Actions (#48985)
Compare 135 commits »
scriptoriumadmin synced commits to refs/pull/48405/head at scriptoriumadmin/litellm from mirror 2026-10-30 06:48:56 +00:00
92a0aad3f0 fix(prometheus): stop litellm_deployment_state sticking at 2 after a cooldown
3f5e42d032 fix(proxy): enforce team and key Allowed Vector Stores when no vector store registry is loaded (#48915)
c666df3dac docs(qa): allow focused proof for mechanical PRs (#49031)
bad98c4b02 fix(responses): call the renamed drops_params helper on the compact path (#49030)
b7da9a2755 fix(credentials): keep the stored secret when PATCH echoes its masked read-back (#48950)
Compare 115 commits »
scriptoriumadmin synced commits to refs/pull/48380/merge at scriptoriumadmin/litellm from mirror 2026-10-30 06:48:55 +00:00
d8f162ea22 fix(guardrails): allow MCP call modes on the generic guardrail API (#48060)
1a2b443a32 ci: run the extensions integration group with 48 workers (#49134)
8d441ebc15 test(otel): give the Langfuse destination tests the otel callback's stdout exporter (#49129)
4ce724e3df fix(caching): key cached async clients by a per-loop token instead of id(loop) (#48877)
Compare 188 commits »