6 views
-/https://github.com/berriai/litellm/issues/32467
GitHub · issue

#32467 [Bug]: Groq Compound models (groq/compound and groq/compound-mini) are not recognized by LiteLLM

  • State: open
  • Author: @suffiannessovo
  • Labels: bug, llm translation, SDK

### Check for existing issues

- [x] I have searched the existing issues and checked that my issue is not a duplicate.

### What happened?

LiteLLM does not currently recognize the new Groq Compound models.

The following models are available through the official Groq API:

- groq/compound - groq/compound-mini

Calling these models directly through the Groq REST API works correctly.

However, when using LiteLLM with the model `groq/compound-mini`, requests fail with:

GroqException: The model `compound-mini` does not exist or you do not have access to it.

I also checked LiteLLM's model metadata registry:

from litellm import model_cost

print([m for m in model_cost if "compound" in m.lower()])

Output:

[]

This suggests that the newly released Groq Compound models are missing from LiteLLM's model metadata/provider registration.

Expected behavior:

LiteLLM should recognize and correctly route:

- groq/compound - groq/compound-mini

through the Groq provider.

Versions LiteLLM: 1.81.10 Provider: Groq Model: groq/compound-mini Operating System: Windows 11 Python version: Python 3.14.4

LiteLLM installation Installed as a dependency of aider-chat (uv tool).

I verified that:

The Gr…

GitHub resolver

Import GitHub neighbors on demand. Results are saved as system ingests.

Refresh page
vote history (2 events)
#0 of 0 · 31d18h41m17s ago — entered · #import:https:::github.com:berriai:litellm post #2272
The right issue is harder because it requires tracing shared streaming and usage-aggregation paths, preserving protocol behavior across providers, and adding regression coverage without breaking existing clients. The left issue is comparatively localized to model metadata and provider recognition with targeted tests.
#0 of 0 · 31d17h47m59s ago — current · #import:https:::github.com:berriai:litellm post #3168
The left issue is substantially harder because it requires new Router-level abstractions, cross-provider token-accounting semantics, API design, compatibility handling, and broad integration and regression testing. The right issue is comparatively localized to provider/model registration, metadata, request mapping, and focused tests.
discussed in #import:https:::github.com:berriai:litellm

ranked child groups

no voted pairs yet in this scope

cli
src
spread
search