I read the 1.7.0 release notes, which highlighted the agent feature, but where exactly is it?

[a86582751]

Is the "Assistant Library" in the interface still exactly the same as the original?

[dqtz5vpvj9-create]

I have kimi and deepseek official providers, but the model list is still empty

[olafgeibig]

Same for me. No models available although I have many configured.

[DeJeune]

restart api Server?

[olafgeibig]

API Server restarted, still no models. I'm on Cherry 1.7.1, MacOS Tahoe 26.1, MacBook Pro M1, 16 GB.

[DeJeune]

API Server restarted, still no models. I'm on Cherry 1.7.1, MacOS Tahoe 26.1, MacBook Pro M1, 16 GB.

Oh. I got it. Currently, this version only supports models provided by Anthropic-type or Anthropic-endpoint providers. Support for any model will be included in the next release.

[olafgeibig]

Interesting... I never used the API server before but something is not okay. It seems to only offer schemas but no methods. I had authenticated with my API key.

[olafgeibig]

For the time being, you can use LiteLLM in proxy mode to create a local anthropic API compatible endpoint pointing to whatever model you want. My example is creating a model anthropic/gpt-5 that actually points to my gpt-5 on Azure. You can install LiteLLM with uv tool install "litellm[proxy]" and then run on your console litellm --config litellm-config.yaml

litellm-config.yaml:

litellm_settings:
  cache: True
  cache_params:
    type: local
model_list:
  - model_name: anthropic/gpt-5
    litellm_params:
      model: azure/gpt5_series/gpt-5
      api_base: https://yourazure.openai.azure.com
      api_version: 2025-04-01-preview
      api_key: "os.environ/AZURE_API_KEY"
    model_info:
      base_model: azure/gpt-5

In Cherry: