Remove deepseek v3.1 terminus

[AnthonyRonning]

We need to remove this model from all documentation and model selecting logic.

Important: Deepseek R1 is staying.

The complicated thing is that we have a new descriptive model selector. v3.1 is an alias for "Reasoning" but with the "brain toggle" turned off.

We should temporarily remove the brain toggle (we'll bring it back at a later time, so comment out / lint ignore works best). The "Reasoning" with "brain off" should now be R1.

Make sure the model is removed from marketing pages and pricing pages too.

[AnthonyRonning]

@claude can you handle this?

[claude]

Claude finished @AnthonyRonning's task in 5m 2s —— View job • claude/issue-343-20251217-1851 • Create PR ➔

---

Successfully removed DeepSeek v3.1 terminus

Completed Tasks

• Read CLAUDE.md for repo guidelines
• Search for DeepSeek v3.1 references in codebase
• Locate model selector logic
• Remove DeepSeek v3.1 from MODEL_CONFIG
• Update CATEGORY_MODELS to use R1 for reasoning_off
• Comment out brain toggle in UnifiedChat.tsx (2 locations)
• Remove v3.1 from marketing page (Marketing.tsx)
• Update pricing page references (already correct)
• Run linting and build checks (all passed!)
• Commit and push changes

Changes Made

1. ModelSelector.tsx:

   • Removed deepseek-v31-terminus from MODEL_CONFIG
   • Updated CATEGORY_MODELS.reasoning_off to use deepseek-r1-0528 instead
   • Added comment noting brain toggle is temporarily disabled

2. UnifiedChat.tsx:

   • Commented out brain toggle button in both locations using false && with proper eslint-disable
   • Code is preserved for easy re-enabling later

3. Marketing.tsx:

   • Removed "DeepSeek V3.1 Terminus" from the AI_MODELS labels array
   • Kept "DeepSeek R1" as requested

All lint checks passed and build succeeded! The "Reasoning" mode now always uses DeepSeek R1, and the brain toggle UI has been temporarily hidden.

--- | Branch: claude/issue-343-20251217-1851