Skip to content

Fix/model provider mismatch#181

Open
vdalal wants to merge 2 commits intovirattt:mainfrom
vdalal:fix/model-provider-mismatch
Open

Fix/model provider mismatch#181
vdalal wants to merge 2 commits intovirattt:mainfrom
vdalal:fix/model-provider-mismatch

Conversation

@vdalal
Copy link
Contributor

@vdalal vdalal commented Mar 14, 2026

Add CLI warning for model-provider mismatch
Tested with: Windows 11

dshark3y and others added 2 commits March 11, 2026 10:48
* Add Gemini 3.1 Pro to Google models

Added Google's 3.1 Pro Model to Google Model Options

* Remove Gemini 3 Pro from Google models

Removed Gemini 3 Pro from the Google models list. It's depreciated. Keeping 3.1 Pro.
@vdalal vdalal force-pushed the fix/model-provider-mismatch branch from 41c7df3 to 41a0021 Compare March 14, 2026 18:30
@virattt
Copy link
Owner

virattt commented Mar 14, 2026

When would this case happen?

@vdalal
Copy link
Contributor Author

vdalal commented Mar 19, 2026

When would this case happen?

Firstly, this is overly defensive. This does not happen during normal user flow using the /model command.

It can happen in out-of-band ways:

  • Manual settings edits: If an advanced user modifies .dexter/settings.json manually and updates "provider": xyz but forgets to update "modelId" (leaving another provider abc's model).
  • Version upgrades/downgrades: If a user updates Dexter, and a previously saved model/provider combination becomes invalid or defaults change incorrectly.

Without warnOnMismatch(), Dexter would try to pass (say) an Anthropic model string to the (say) OpenAI client API, which would result in an obscure LLM API error (like "model claude-3-haiku does not exist"). This check just gives the user a clear, actionable CLI error instead.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants