The wrong context window is being reported for OpenRouter models for specific providers

[daniel-lxs]

App Version

3.16.4

API Provider

OpenRouter

Model Used

Llama 4 Scout

Actual vs. Expected Behavior

Current Behavior:

• src/api/providers/openrouter.ts uses rawModel.context_length (model-level context window)
• This causes incorrect context window display when different providers support different sizes for the same model

Expected Behavior:

• Should use provider-specific context window size (endpoint.context_length)
• As correctly implemented in webview-ui/src/components/ui/hooks/useOpenRouterModelProviders.ts

Impact:

• Leads to incorrect token limit calculations
• May cause unexpected behavior when provider's actual context window is smaller than displayed

Detailed Steps to Reproduce

Reproduction Steps:

1. Create an OpenRouter profile
2. Select Llama 4 Scout model
3. Choose Cerebras provider (32k tokens context window)
4. Observe task header shows 1M tokens (model default) instead of 32k