Skip to main content
Some opinionated advice on which model to pick for common use cases.

​
Speed vs quality vs price

​
Long context

Need >100k tokens? Sort by maxContextTokens:
  • qwen-long — 10M tokens
  • grok-4.20 — 2M tokens
  • gemini-2.5-pro / flash — 1M tokens
  • claude-opus-4.7 / 4.6 / sonnet-4.6 — 1M tokens
  • qwen3.5-plus / flash — 1M tokens
  • gpt-5.5 — 272K tokens
  • gpt-5.4 / 5.2 — 200K tokens
  • gpt-4o — 128K tokens

​
Vision

For images in prompts:
  • General-purpose: gpt-4o (best accuracy on OCR + natural images)
  • Huge context (many images): gemini-2.5-pro
  • Document-heavy (PDF + text): claude-opus-4.7 or qwen3-max

​
Coding

  • claude-opus-4.7 — step-change improvement in agentic coding over 4.6
  • claude-sonnet-4.6 — fast, great at code review and refactor
  • gpt-5.5 / 5.4 — strong on general programming
  • deepseek-v3.1 / qwen3-coder-plus — budget options for mass code gen

​
Tool calling

All major flagships (gpt-5.5, gpt-5.4, claude-opus-4.7, gemini-2.5-pro, grok-4.20, qwen3-max, deepseek-v3.1) support tools. For best-reliability production agents, pick one and stick to it — don’t round-robin.

​
JSON mode / Structured Outputs

  • gpt-5.5 / 5.4 / o3 — best at strict JSON schema
  • gemini-2.5-pro — excellent
  • claude-opus-4.7 — good; returns strict JSON when requested

​
Cost-sensitive

The cheap tier moves too fast for a table here to stay true — a price list in static documentation is a promise about money that nobody is maintaining. Sort the live catalogue by price, or read pricing from GET /v1/models and pick in code. As a shape rather than a number: the budget tier sits roughly an order of magnitude below the flagships, and the small models from Alibaba, Google and DeepSeek are usually the floor.

​
Fallback pairs

Safe production pairs (primary → fallback):
  • gpt-5.5 → claude-opus-4.7 (cross-provider top-tier)
  • gpt-5.5 → gpt-5.4 (within OpenAI, cheaper)
  • gpt-5.4 → gpt-5.4-mini
  • claude-opus-4.7 → gpt-5.5 (cross-provider top-tier)
  • claude-opus-4.7 → claude-sonnet-4.6
  • gemini-2.5-pro → gemini-2.5-flash
  • gpt-5.4 → gemini-2.5-flash (cross-provider safety)
Configure in Settings → Fallbacks. See Fallback chains.