scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-09-16 23:41:43 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 103

fix: messages and responses are supported — measured, not assumed

Browse source 
I first declared both false on the grounds that I had not tested LiteLLM's bridge
against this endpoint. That was the wrong way to resolve it; I ran the calls instead.

Against litellm 1.97.0 with a live key:

  litellm.responses(...).output_text            -> 'ok'
  await litellm.anthropic_messages(...)         -> Anthropic-shaped content blocks,
                                                   with reasoning mapped into a
                                                   'thinking' block
  litellm.completion(..., stream=True)          -> content plus 110 reasoning_content
                                                   chunks
  litellm.text_completion(...)                  -> 'Paris.'
  tool_calls, response_format json_schema       -> both correct

Everything left false was probed too, rather than assumed: embeddings and moderations
fall through to another provider's credentials, audio and rerank raise 'tiyuvta is not
a valid LlmProviders', batches needs a files API this endpoint does not have, and
litellm 1.97.0 exposes no a2a entry point.
This commit is contained in:
Avi Fenesh 2026-08-17 13:15:51 +03:00
parent 3353a5656d
commit eab6281923
1 changed files with 2 additions and 2 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

4
provider_endpoints_support.json
View file

@ -2371,8 +2371,8 @@
"url": "https://inference.tiyuvta.ai/docs",
"endpoints": {
"chat_completions": true,
"messages": false,
"responses": false,
"messages": true,
"responses": true,
"embeddings": false,
"image_generations": false,
"audio_transcriptions": false,