scriptoriumadmin
  • Joined on 2026-03-01
scriptoriumadmin synced commits to refs/pull/40975/merge at scriptoriumadmin/litellm from mirror 2026-10-29 06:35:20 +00:00
a66dc50efb test(json_fragment_accumulator): assert peak allocation instead of a drain timing ratio (#43370)
cbf4e40d38 fix(proxy)!: stop labelling management and health rejections as auth failures (#48739)
a11e76fe8f fix(google_genai): read one-object and null spellings of repeated fields in countTokens and generateContent (#48771)
764a7a6536 fix(context_management)!: count the system prompt toward the clear_tool_uses input_tokens trigger (#47857)
Compare 216 commits »
scriptoriumadmin synced commits to refs/pull/40969/merge at scriptoriumadmin/litellm from mirror 2026-10-29 06:35:20 +00:00
23b418eb48 feat(prompt_templates): opt-in LITELLM_DROP_MIDTURN_SYSTEM removes system turns after the leading run
cbf4e40d38 fix(proxy)!: stop labelling management and health rejections as auth failures (#48739)
a11e76fe8f fix(google_genai): read one-object and null spellings of repeated fields in countTokens and generateContent (#48771)
764a7a6536 fix(context_management)!: count the system prompt toward the clear_tool_uses input_tokens trigger (#47857)
Compare 227 commits »
scriptoriumadmin synced commits to refs/pull/40954/merge at scriptoriumadmin/litellm from mirror 2026-10-29 06:35:19 +00:00
4fc2bba1f2 fix(stream): the last usage frame carries litellm's bill in usage.cost (#48955)
019a9de603 fix(logging): stop dumping the request body at every layer under --detailed_debug (#48973)
57f05e56f6 fix(bedrock): omit system prompt from audio transcription and report its token usage (#41565)
d9c9d84298 fix(proxy): stop log lines from leaving terminal colors on (#48966)
Compare 210 commits »
scriptoriumadmin synced commits to refs/pull/40950/merge at scriptoriumadmin/litellm from mirror 2026-10-29 06:35:19 +00:00
ce9dc9fd48 fix(streaming): don't treat usage.cost=0 as Vertex stream spend
3b9a59348f fix(vertex_ai)!: send the caller's anthropic-version in the Vertex Claude /v1/messages body, not the header (#48814)
07d314beb5 test(responses): audit #47445 (#48902)
77cad6e903 test(proxy): audit #46997 (#48918)
Compare 2335 commits »
scriptoriumadmin synced commits to refs/pull/40967/head at scriptoriumadmin/litellm from mirror 2026-10-29 06:35:19 +00:00
69e5741e5b fix(anthropic): accept server_tool_use and web_search_tool_result blocks when costing a /v1/messages response
b859137b24 Merge remote-tracking branch 'origin/main' into fix_fake_stream_server_tool_blocks
aa769fbc69 fix(streaming): surface upstream abort as server errors (LIT-10159) (#48328)
1f68cc5649 refactor(rust-bridge): require the native extension (#48852)
99ac0d3e39 fix(aiohttp_openai): forward service_tier in transform_response (#47263)
Compare 761 commits »
scriptoriumadmin synced commits to refs/pull/40966/merge at scriptoriumadmin/litellm from mirror 2026-10-29 06:35:19 +00:00
2cddcf31b2 feat(chatgpt): opt-in stable session id derived from the conversation prefix
cbf4e40d38 fix(proxy)!: stop labelling management and health rejections as auth failures (#48739)
a11e76fe8f fix(google_genai): read one-object and null spellings of repeated fields in countTokens and generateContent (#48771)
764a7a6536 fix(context_management)!: count the system prompt toward the clear_tool_uses input_tokens trigger (#47857)
Compare 286 commits »
scriptoriumadmin synced commits to refs/pull/40966/head at scriptoriumadmin/litellm from mirror 2026-10-29 06:35:19 +00:00
2cddcf31b2 feat(chatgpt): opt-in stable session id derived from the conversation prefix
cbf4e40d38 fix(proxy)!: stop labelling management and health rejections as auth failures (#48739)
a11e76fe8f fix(google_genai): read one-object and null spellings of repeated fields in countTokens and generateContent (#48771)
764a7a6536 fix(context_management)!: count the system prompt toward the clear_tool_uses input_tokens trigger (#47857)
de42b67252 fix(proxy)!: enforce an end user's per-model budget from their linked budget (#48815)
Compare 2304 commits »
scriptoriumadmin synced commits to refs/pull/40965/merge at scriptoriumadmin/litellm from mirror 2026-10-29 06:35:19 +00:00
2e258547ee fix(chatgpt): keep structured output text.format in Responses API requests
cbf4e40d38 fix(proxy)!: stop labelling management and health rejections as auth failures (#48739)
a11e76fe8f fix(google_genai): read one-object and null spellings of repeated fields in countTokens and generateContent (#48771)
764a7a6536 fix(context_management)!: count the system prompt toward the clear_tool_uses input_tokens trigger (#47857)
Compare 343 commits »
scriptoriumadmin synced commits to refs/pull/40950/head at scriptoriumadmin/litellm from mirror 2026-10-29 06:35:19 +00:00
ce9dc9fd48 fix(streaming): don't treat usage.cost=0 as Vertex stream spend
3b9a59348f fix(vertex_ai)!: send the caller's anthropic-version in the Vertex Claude /v1/messages body, not the header (#48814)
07d314beb5 test(responses): audit #47445 (#48902)
77cad6e903 test(proxy): audit #46997 (#48918)
e1580b7b83 refactor(images): remove dall-e-2 and dall-e-3 code paths (#48703)
Compare 6609 commits »
scriptoriumadmin synced commits to refs/pull/40965/head at scriptoriumadmin/litellm from mirror 2026-10-29 06:35:19 +00:00
2e258547ee fix(chatgpt): keep structured output text.format in Responses API requests
cbf4e40d38 fix(proxy)!: stop labelling management and health rejections as auth failures (#48739)
a11e76fe8f fix(google_genai): read one-object and null spellings of repeated fields in countTokens and generateContent (#48771)
764a7a6536 fix(context_management)!: count the system prompt toward the clear_tool_uses input_tokens trigger (#47857)
de42b67252 fix(proxy)!: enforce an end user's per-model budget from their linked budget (#48815)
Compare 822 commits »
scriptoriumadmin synced commits to refs/pull/40631/merge at scriptoriumadmin/litellm from mirror 2026-10-29 06:35:18 +00:00
48e93ab92f fix(http2): read an unquoted http2: 0 as off and free the stream slot of a response dropped without closing
764a7a6536 fix(context_management)!: count the system prompt toward the clear_tool_uses input_tokens trigger (#47857)
de42b67252 fix(proxy)!: enforce an end user's per-model budget from their linked budget (#48815)
5bd47442fb fix(cli): give lite codex aliases of a Codex-known model that model's catalog entry (#48894)
Compare 389 commits »
scriptoriumadmin synced commits to refs/pull/40810/head at scriptoriumadmin/litellm from mirror 2026-10-29 06:35:18 +00:00
0c327e5f63 chore(gitgot): regenerate provider_endpoints_support.json from the cost map
c1033a41c0 Merge branch 'main' into add-gitgot-provider
e32d8a4d03 fix(responses): reject or drop Responses params the chat completions bridge cannot translate (#48655)
df9091deb6 fix(responses): skip failed spend-log turns when rebuilding a session from previous_response_id (#48647)
c7c895db45 fix(proxy)!: paginate customer lists and usage filters (#48700)
Compare 669 commits »
scriptoriumadmin synced commits to refs/pull/40810/merge at scriptoriumadmin/litellm from mirror 2026-10-29 06:35:18 +00:00
83ca14b392 fix(pricing): sync databricks model prices (#48884)
1bd9a0780a refactor(ui): use the shadcn Kbd primitive for keyboard hints (#48885)
7c7304c22b feat(providers): add ZeroGPU as an OpenAI-compatible provider (#43874)
ca6c830221 fix(proxy): account for media in combined TPM reservations (#48865)
Compare 613 commits »
scriptoriumadmin synced commits to refs/pull/40701/merge at scriptoriumadmin/litellm from mirror 2026-10-29 06:35:18 +00:00
4bbc2c3e2d fix: satisfy typing for silent model fanout
ca6c830221 fix(proxy): account for media in combined TPM reservations (#48865)
9675c13cb0 fix(openai)!: send the whole reasoning_effort dict when chat is bridged to Responses (#48807)
41fcce62ff fix(gemini)!: accept Google's current Managed Agents fields on create and interactions (#48750)
Compare 161 commits »
scriptoriumadmin synced commits to refs/pull/40701/head at scriptoriumadmin/litellm from mirror 2026-10-29 06:35:18 +00:00
4bbc2c3e2d fix: satisfy typing for silent model fanout
483e70552c merge: integrate main while preserving silent model guards
6698ce0745 fix: mirror generic requests to every silent model with isolated metadata
1176e608b1 test(harness): audit #47530 (#48751)
220a70653e docs(readme): link the LiteLLM OSS Champions program from Contributing (#48734)
Compare 157 commits »
scriptoriumadmin synced commits to refs/pull/40631/head at scriptoriumadmin/litellm from mirror 2026-10-29 06:35:18 +00:00
48e93ab92f fix(http2): read an unquoted http2: 0 as off and free the stream slot of a response dropped without closing
b79bf68f8a test(http2): move the HTTP/2 eviction cases that open a loopback server into the integration wire tests
6813be09fd fix(http2): count proxy-mounted transports when checking an evicted client for in-flight requests
4401833140 fix(http2): keep TE, force_ipv4 trust, evicted-client and opt-out behavior on the HTTP/2 path
939b86f6ae test(http2): scope wire-test peer logs to each fixture instance
Compare 454 commits »
scriptoriumadmin synced commits to refs/pull/40457/merge at scriptoriumadmin/litellm from mirror 2026-10-29 06:35:18 +00:00
d9c9d84298 fix(proxy): stop log lines from leaving terminal colors on (#48966)
7aa7948a47 fix(health): ask OpenAI and Azure audio chat models for audio output in the health check probe (#47730)
e4932a4539 test(router): set string fallback targets after Router init in the sdk cooldown cases (#48954)
57c1df8c33 refactor(model-catalog): parse, lower and validate the catalog into typed prices (#48853)
Compare 157 commits »
scriptoriumadmin synced commits to refs/pull/40683/merge at scriptoriumadmin/litellm from mirror 2026-10-29 06:35:18 +00:00
38b4d31447 test(scim): audit #47544 (#48791)
356069d2ed test(mcp): audit #47422 (#48821)
3507781735 test(proxy): audit #47352 (#48804)
295d79fc7d feat(cli): keep the ChatGPT login with lite codex --subscription (#48741)
Compare 115 commits »
scriptoriumadmin synced commits to refs/pull/40264/merge at scriptoriumadmin/litellm from mirror 2026-10-29 06:35:18 +00:00
cbf4e40d38 fix(proxy)!: stop labelling management and health rejections as auth failures (#48739)
a11e76fe8f fix(google_genai): read one-object and null spellings of repeated fields in countTokens and generateContent (#48771)
764a7a6536 fix(context_management)!: count the system prompt toward the clear_tool_uses input_tokens trigger (#47857)
de42b67252 fix(proxy)!: enforce an end user's per-model budget from their linked budget (#48815)
Compare 215 commits »
scriptoriumadmin synced commits to refs/pull/40726/merge at scriptoriumadmin/litellm from mirror 2026-10-29 06:35:18 +00:00
43ce3e22aa fix(responses)!: answer 400 before counting a lone Responses built-in tool on Claude (#48798)
ff5bf4826d fix(count_tokens)!: audit #47441 and keep the framed estimate for Bedrock count models from the cost map (#48805)
9b1d69e32c fix(images): normalize absent images in async edits (#48872)
7abff3dd63 fix(mistral): replay assistant reasoning as a Mistral thinking chunk (#48778)
Compare 137 commits »