Claude Series

Anthropic Claude series excels at long-form content and coding, making it the top choice for writing and programming scenarios.

Available Models

Model	Positioning
claude-sonnet-5	Current mainstay, balanced quality/cost
claude-opus-4-6	Flagship, higher ceiling for complex tasks
claude-opus-4-6-thinking	Flagship thinking tier, reasoning for hard problems
claude-sonnet-4-6(-thinking)	Previous mainstay (including thinking tier)
claude-haiku-4-5	Lightweight tier, low-cost batch processing

For real-time pricing, see Model Marketplace.

Usage

Choose either protocol:

• OpenAI-compatible (recommended, zero changes): set model to a Claude model name; see Compatible Mode Calls;
• Anthropic native: full features (cache_control, thinking parameters), see Claude Native Calls.

Selection Recommendations

• Long-form writing / revision: claude-sonnet-5, stable style control;
• Coding assistant: claude-sonnet-5 for daily use, claude-opus-4-6 for difficult tasks, tool integration see Claude Code;
• Auditing / grading / math: claude-opus-4-6-thinking, usage see Thinking Mode;
• Customer service / batch processing: claude-haiku-4-5;
• For long system prompts, be sure to use Claude Prompt Caching to reduce costs.