Skip to content
Clipboard Recast

Vision Models

Vision Models

Section titled “Vision Models”

Not all AI models can analyze images and screenshots. This guide shows which models support vision and how to configure them.

Vision-Capable Providers

Section titled “Vision-Capable Providers”

✅ Claude (Anthropic) - RECOMMENDED

Section titled “✅ Claude (Anthropic) - RECOMMENDED”

All Claude 3+ models support vision by default.

Models with Vision:

  • ✅ claude-sonnet-4-5 (Default) - Excellent
  • ✅ claude-3-opus-20240229 - Best quality
  • ✅ claude-3-sonnet-20240229 - Good
  • ✅ claude-3-haiku-20240307 - Fast, cheap

Best For:

  • Code screenshots
  • UI/UX analysis
  • Error debugging
  • General screenshot analysis

Setup: Default model works perfectly - no changes needed!

✅ OpenAI - EXCELLENT

Section titled “✅ OpenAI - EXCELLENT”

Vision-specific models required.

Models with Vision:

  • ✅ gpt-4-vision-preview - Recommended
  • ✅ gpt-4-turbo - Includes vision
  • ✅ gpt-4o - Optimized multimodal
  • ✅ gpt-4o-mini - Budget option

Best For:

  • Detailed descriptions
  • OCR (text extraction)
  • Chart/graph analysis

Setup: Change model to gpt-4-vision-preview in Settings

✅ Google Gemini - GOOD

Section titled “✅ Google Gemini - GOOD”

Requires vision-specific model.

Models with Vision:

  • ✅ gemini-pro-vision - Must use this!
  • ✅ gemini-1.5-pro - Latest
  • ✅ gemini-1.5-flash - Faster

Best For:

  • Charts and graphs
  • Scientific diagrams
  • Free tier testing

Setup: ⚠️ IMPORTANT - Change model from gemini-pro to gemini-pro-vision in Settings

✅ OpenRouter - FLEXIBLE

Section titled “✅ OpenRouter - FLEXIBLE”

Many vision models available.

Models with Vision:

  • ✅ anthropic/claude-sonnet-4-5 (Default) - Excellent
  • ✅ openai/gpt-4-vision-preview
  • ✅ google/gemini-pro-vision

Setup: Default model (Claude) works great

❌ No Vision Support

Section titled “❌ No Vision Support”

These providers cannot analyze screenshots:

  • ❌ Cerebras
  • ❌ Together AI
  • ❌ z.ai

Quick Setup Guide

Section titled “Quick Setup Guide”

For Claude (Easiest)

Section titled “For Claude (Easiest)”
Provider: Claude (Anthropic)
Model: claude-sonnet-4-5 (default is fine)

No changes needed - works out of the box!

For OpenAI

Section titled “For OpenAI”
Provider: OpenAI
Model: gpt-4-vision-preview (change from default)

Must change model for vision support.

For Gemini (Free Tier)

Section titled “For Gemini (Free Tier)”
Provider: Google Gemini
Model: gemini-pro-vision (MUST CHANGE from gemini-pro)

Critical: Default model doesn’t support vision!

Use Case Recommendations

Section titled “Use Case Recommendations”

Code Screenshots

Section titled “Code Screenshots”

Best: Claude 3.5 Sonnet

  • Understands syntax highlighting
  • Explains code logic well

UI/UX Design

Section titled “UI/UX Design”

Best: Claude or GPT-4 Vision

  • Great design critique
  • Understands layouts

Error Debugging

Section titled “Error Debugging”

Best: Claude 3.5 Sonnet

  • Understands stack traces
  • Suggests solutions quickly

Charts & Graphs

Section titled “Charts & Graphs”

Best: Gemini Pro Vision or GPT-4 Vision

  • Excellent at data interpretation
  • Extracts numbers accurately

Budget Option

Section titled “Budget Option”

Best: Claude Haiku or GPT-4o Mini

  • Much cheaper
  • Still decent quality

Testing Your Setup

Section titled “Testing Your Setup”

After configuring:

  1. Take a screenshot (Cmd+Shift+4)
  2. Wait 2 seconds
  3. Press Cmd+Shift+A
  4. Select “Explain”
  5. You should see a description of your screenshot

If It Doesn’t Work

Section titled “If It Doesn’t Work”

Error: “Model doesn’t support images”

  • Using wrong model (e.g., gemini-pro instead of gemini-pro-vision)
  • Change model in Settings

Error: “Failed to process”

  • API key might not have access to vision models
  • Provider doesn’t support vision
  • Check provider status page

Cost Comparison

Section titled “Cost Comparison”

Claude Models

Section titled “Claude Models”
  • Sonnet 4.5: $3-15 per million tokens
  • Opus: $15-75 per million tokens
  • Haiku: $0.25-1.25 per million tokens

OpenAI Models

Section titled “OpenAI Models”
  • GPT-4 Vision: $10-30 per million tokens
  • GPT-4o: $5-15 per million tokens
  • GPT-4o Mini: $0.15-0.60 per million tokens

Gemini Models

Section titled “Gemini Models”
  • Free tier: 60 requests/minute
  • Paid: ~$0.50 per million tokens

Best Overall Setup

Section titled “Best Overall Setup”

For most users:

Provider: Claude (Anthropic)
Model: claude-sonnet-4-5
Why: Best balance of quality, speed, and cost
No configuration needed!

Next Steps

Section titled “Next Steps”