Model for inline edit

[sr-tream]

Please describe the feature you want

Split chat and inline edit model.

Motivation:

1. inline edit do not support thinking models
2. inline edits can be done with smaller models, what working faster

Additional context

Screenshot is demonstrated current state with inline edit, when used reasoning model. Reasoning models is very good for chat, but bad for inline editing

Workaround for current version

Add non-thinking model to supported_models and change model before inline editing. Example:

[model.chat.http]
kind = "openai/chat"
model_name = "qwen2.5-coder:3b"
api_endpoint = "http://127.0.0.1:11434/v1"
supported_models = ["qwen3", "rhundt/qwen3-64k:30b", "qwen2.5-coder:3b"]

NOTE: model_name is used for inline edit here, but as default used first model from supported_models.

---

Please reply with a 👍 if you want this feature.

[zwpaper]

Certainly, thank you for the report. The issue with reasoning models is already on our radar. We will investigate it further in due course. In the meantime, please feel free to utilize the workaround as you've outlined.