anthropic
Anthropic is the developer of the Claude family of large language models, with a focus on AI safety and alignment research. Claude is renowned for its long context, strong reasoning, and coding and tool-use capabilities, spanning three sizes: Opus (flagship), Sonnet (recommended), and Haiku (ultra-fast). It natively supports a 200K context window, with select models extending to 1M.
200K–1M token context window; prompt caching (cache_read 10%, cache_creation 1.25×); visual understanding (images/PDF); tool calling and Computer Use; Extended Thinking; MCP compatibility; JSON mode; system prompt customization; Batch API at half price.
Code generation and refactoring; long-document analysis (PDF / codebases); multi-turn agent tasks; enterprise knowledge base Q&A; chart data extraction; automated office operations; security review and compliance auditing.
No image generation; rate limits are based on account quotas; no on-premises deployment; maximum output length is bound by each model's max_tokens; some enterprise features require contacting sales to enable.
| Model | Input / 1M | Output / 1M | Cache Read / 1M | Per 1M |
|---|---|---|---|---|
| 1M | ||||
| 1M | ||||
| 1M | ||||
| 1M | ||||
| 1M | ||||
| $15/M$10.5/M-30% |
| $0.3/M$0.21/M-30% |
| 1M |
Claude Opus 4.6 claude-opus-4-6 | $5/M$3.5/M-30% | $25/M$17.5/M-30% | $0.5/M$0.35/M-30% | 1M |
Claude Opus 4.5 claude-opus-4-5 | $5/M$3.5/M-30% | $25/M$17.5/M-30% | $0.5/M$0.35/M-30% | 1M |
Claude Haiku 4.5 claude-haiku-4-5 | $1/M$0.7/M-30% | $5/M$3.5/M-30% | $0.1/M$0.07/M-30% | 1M |
Claude Sonnet 4.5 claude-sonnet-4-5 | $3/M$2.1/M-30% | $15/M$10.5/M-30% | $0.3/M$0.21/M-30% | 1M |