Claude Opus 4
by Anthropic
World's best coding model released May 2025, leading SWE-bench at 72.5% and Terminal-bench at 43.2%. Designed for sustained performance on complex long-running tasks requiring thousands of steps and several hours of continuous work.
- World's best coding model with sustained performance on complex tasks requiring thousands of steps
- Continuous operation capability working for several hours on long-running agent workflows
- Leading performance on SWE-bench (72.5%) and Terminal-bench (43.2%) coding benchmarks
- Enhanced memory capabilities with improved file access and tacit knowledge building
- Frontier agent performance powering complex codebase understanding and autonomous development
Web Search
Search the web for current info.
Extended Thinking
Enhanced reasoning for complex tasks.
Web Fetch
Fetch and read full content from any URL.
Code Execution
Run Python & Bash code in a sandbox.
JSON Mode
Output responses in valid JSON format.
Model Information
Supported Formats
Claude Fable 5Anthropic's most capable widely released model, for the most demanding reasoning and long-horizon agentic work. Thinking is always on with a 1M token context window (also the maximum) and 128K max output. Priced above the Opus tier for the highest capability ceiling in the Claude lineup.
Claude Opus 5Anthropic's successor to Opus 4.8 for complex agentic coding and enterprise work. Strongest on deep reasoning, agentic and long-horizon work, and test-time compute scaling, at half the cost of Claude Fable 5. Thinking is on by default with a 1M token context window and 128K max output.
Claude Sonnet 5Best combination of speed and intelligence in the Sonnet tier, delivering near-Opus quality on coding and agentic work. Successor to Sonnet 4.6 with adaptive thinking on by default, a 1M token context window, and 128K max output.