scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-08-28 05:25:59 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 36
litellm/litellm
History
Exact
tin-berri 3300fc3a96
fix(moonshot, together_ai): send the reasoning effort Kimi K3 accepts (#38611) 
* fix(moonshot, together_ai): send the reasoning effort Kimi K3 accepts

Moonshot documents reasoning_effort as a top-level chat completions field for its reasoning
models, and defaults it to max, but MoonshotChatConfig builds its supported params by
subtracting from the OpenAI base list, which never carried that param. An explicit level
raised UnsupportedParamsError before the request left the proxy, so low and high were
unreachable and every call ran at the provider default

Together accepts low, high and max on Kimi K3. The per-model clamp added for the gpt-oss
family folds max down to high for every model except deepseek-ai/DeepSeek-V4-Pro, so a caller
asking for max silently got roughly half the reasoning budget they paid for

Moonshot now offers reasoning_effort whenever the registry says the model reasons. Together
sends a level the map entry declares unchanged, and keeps its existing table for every level
an entry does not name, so the only value that moves is Kimi K3 at max

* fix(moonshot): unwrap the bridges' effort object to the level string
2026-08-27 20:46:46 -07:00
..
a2a_protocol fix(a2a): speak the 0.3 dialect to servers with mis-cased protocol bindings 2026-08-24 10:51:14 -07:00
anthropic_interface chore(lint): strip inert type: ignore comments and zero LIT009, LIT010, LIT011 headroom 2026-08-05 02:37:24 -07:00
assistants chore(lint): clear grandfathered over-limit lint drift and ratchet budgets down 2026-08-05 12:18:13 -07:00
batch_completion
batches chore(typing): tighten annotations added in the last day and ratchet budgets 2026-08-23 07:49:54 +00:00
caching fix(caching): treat an empty cache namespace as no namespace 2026-08-26 13:48:45 -07:00
completion_extras fix(anthropic): carry tool_reference tool results through the guardrail translation round trip 2026-08-26 22:35:46 -07:00
compression chore(lint): strip inert type: ignore comments and zero LIT009, LIT010, LIT011 headroom 2026-08-05 02:37:24 -07:00
containers chore(typing): drop 1.3k basedpyright errors across 30 Any hotspot files 2026-08-16 03:56:02 +00:00
endpoints/speech/speech_to_completion_bridge fix(speech): keep proxy metadata and completion cost through the TTS completion bridge 2026-08-26 15:12:41 -07:00
evals chore(typing): remove 914 basedpyright Any errors across 16 hotspot files 2026-08-10 01:24:40 -07:00
experimental_mcp_client feat(mcp): let a resolved OAuth token target a custom upstream header (#38456) 2026-08-27 14:32:01 -07:00
files fix(batches): accept litellm_proxy in files and batches provider type literals 2026-08-18 13:28:06 -07:00
fine_tuning Merge branch 'litellm_internal_staging' into litellm_remove_types_ruff_exclusion 2026-08-05 11:05:33 -07:00
google_genai refactor(google_genai): pick the stream logging endpoint type at construction 2026-08-26 17:18:44 -07:00
images Merge origin/litellm_internal_staging into litellm_fix_36493_image_video_routes 2026-08-24 11:58:22 -07:00
integrations fix(shadow_eval): refuse a judge model that also serves one of the arms it grades (#38589) 2026-08-27 18:44:44 -07:00
interactions feat(vertex_ai): add native Vertex AI Interactions API support 2026-08-25 09:55:13 -07:00
litellm_core_utils fix(shadow_eval): refuse a judge model that also serves one of the arms it grades (#38589) 2026-08-27 18:44:44 -07:00
llms fix(moonshot, together_ai): send the reasoning effort Kimi K3 accepts (#38611) 2026-08-27 20:46:46 -07:00
models refactor(repositories): type prisma table access with one generic protocol 2026-08-25 12:14:17 +00:00
ocr fix(ocr): reject invalid req_format values as 400 on the SDK path 2026-08-17 15:49:18 -07:00
passthrough fix(bedrock_mantle): register a Bedrock runtime passthrough config so /bedrock/model/<deployment>/invoke works 2026-08-25 10:10:38 -07:00
proxy fix(shadow_eval): refuse a judge model that also serves one of the arms it grades (#38589) 2026-08-27 18:44:44 -07:00
proxy_auth
rag chore(typing): remove 914 basedpyright Any errors across 16 hotspot files 2026-08-10 01:24:40 -07:00
realtime_api fix(health): probe Azure GA realtime path for transcription-only models 2026-08-26 12:14:52 -07:00
repositories feat(proxy): opt-in budget rollover carrying overage into the next window (#38514) 2026-08-27 12:46:09 -07:00
rerank_api fix(together_ai): default endpoints to api.together.ai instead of api.together.xyz 2026-08-25 10:18:55 -07:00
responses fix(realtime): carry audio output tokens into response.done usage so Gemini Live native audio bills at the audio rate 2026-08-26 21:10:12 -07:00
router_strategy fix(router): reject complexity-router settings written outside complexity_router_config (#38570) 2026-08-27 17:04:49 -07:00
router_utils fix(moonshot, together_ai): send the reasoning effort Kimi K3 accepts (#38611) 2026-08-27 20:46:46 -07:00
rust_bridge feat(rust): route /chat/completions through the Rust core for anthropic and bedrock (#37241) 2026-08-20 16:15:24 -07:00
sandbox
search chore(lint): strip inert type: ignore comments and zero LIT009, LIT010, LIT011 headroom 2026-08-05 02:37:24 -07:00
secret_managers fix(azure): prefer workload identity over managed identity 2026-08-20 13:45:56 -07:00
skills chore(lint): strip inert type: ignore comments and zero LIT009, LIT010, LIT011 headroom 2026-08-05 02:37:24 -07:00
types feat(proxy): dry-run a real request body on /auto_router/test_routing (#38590) 2026-08-28 01:35:10 +00:00
vector_store_files chore(lint): clear grandfathered over-limit lint drift and ratchet budgets down 2026-08-05 12:18:13 -07:00
vector_stores refactor(repositories): type prisma table access with one generic protocol 2026-08-25 12:14:17 +00:00
videos refactor(videos): make video upload param keyword-only on public edit fns 2026-08-24 15:55:07 -07:00
__init__.py feat(proxy): opt-in budget rollover carrying overage into the next window (#38514) 2026-08-27 12:46:09 -07:00
_internal_context.py
_lazy_imports.py fix(lint): bring basedpyright rule counts back under their budget limits 2026-08-05 10:23:02 -07:00
_lazy_imports_registry.py fix(together_ai): route chat completions through a dedicated TogetherAIChatConfig 2026-08-25 11:44:06 -07:00
_logging.py fix(logging): stop stream-based log collectors classifying INFO logs as errors (#38476) 2026-08-27 17:42:00 -07:00
_redis.py refactor(redis): drop docstrings restating the code 2026-08-26 10:45:56 -07:00
_redis_credential_provider.py chore(lint): strip inert type: ignore comments and zero LIT009, LIT010, LIT011 headroom 2026-08-05 02:37:24 -07:00
_service_logger.py chore(lint): clear grandfathered over-limit lint drift and ratchet budgets down 2026-08-05 12:18:13 -07:00
_uuid.py chore(lint): strip inert type: ignore comments and zero LIT009, LIT010, LIT011 headroom 2026-08-05 02:37:24 -07:00
_version.py
anthropic_beta_headers_config.json
anthropic_beta_headers_manager.py
blog_posts.json
budget_manager.py
constants.py fix(presidio): chunk oversized text before /analyze so large content blocks do not fail (#38483) 2026-08-27 17:41:55 -07:00
cost.json
cost_calculator.py Merge remote-tracking branch 'origin/litellm_internal_staging' into litellm_together_cache_pricing 2026-08-27 14:29:46 -07:00
exceptions.py feat(proxy): redact or drop individual batch records instead of rejecting the file (#37561) 2026-08-20 13:12:55 -07:00
main.py fix: suppress misleading register_model unresolved-cost warnings for entries without custom pricing (#38542) 2026-08-27 12:45:30 -07:00
model_prices_and_context_window_backup.json Merge pull request #38100 from FelipeRodriguesGare/bugfix/tencent-thinking-extra-body 2026-08-27 18:23:11 -07:00
policy_templates_backup.json
provider_endpoints_support_backup.json fix(scx-ai): use the published scx.ai rates and the scx_ai docs url 2026-08-21 17:28:12 +10:00
py.typed
router.py Merge pull request #38606 from BerriAI/litellm_bedrock_messages_midstream_fallback 2026-08-27 19:13:18 -07:00
scheduler.py
setup_wizard.py chore(lint): strip inert type: ignore comments and zero LIT009, LIT010, LIT011 headroom 2026-08-05 02:37:24 -07:00
timeout.py
utils.py feat(model_prices): let a map entry declare its exact reasoning_effort levels (#38481) 2026-08-27 15:38:01 -07:00