Bug: No warning when selected provider/model does not support vision
Environment
- OS: Windows 11
- Provider: Groq
- Model: llama-3.3-70b-versatile
Description
When using a provider or model that does not support vision (e.g. Groq with llama-3.3-70b-versatile), and includeScreenshotByDefault is enabled, the app silently fails with a cryptic error instead of warning the user.
Error received
Groq error 400: messages[2].content must be a string
Root cause
Groq does not support multimodal content (image + text arrays). The app sends screenshots regardless of whether the provider supports vision, causing every message to fail.
Expected behavior
The app should detect when the selected provider/model does not support vision and either:
- Automatically disable screenshots for that provider
- Or show a clear warning in Settings: "This provider/model does not support vision. Screenshots will be disabled."
Workaround (until fixed)
Manually set includeScreenshotByDefault to false in the config file:
powershell -Command "(Get-Content '%APPDATA%\OpenGuider\config.json') -replace '\"includeScreenshotByDefault\": true', '\"includeScreenshotByDefault\": false' | Set-Content '%APPDATA%\OpenGuider\config.json'"
Bug: No warning when selected provider/model does not support vision
Environment
Description
When using a provider or model that does not support vision (e.g. Groq with llama-3.3-70b-versatile), and
includeScreenshotByDefaultis enabled, the app silently fails with a cryptic error instead of warning the user.Error received
Root cause
Groq does not support multimodal content (image + text arrays). The app sends screenshots regardless of whether the provider supports vision, causing every message to fail.
Expected behavior
The app should detect when the selected provider/model does not support vision and either:
Workaround (until fixed)
Manually set
includeScreenshotByDefaulttofalsein the config file: