claude-opus-4-7

Claude Opus 4.7 is the follow-up version in the Opus high-power series, suitable for complex analysis, deep code tasks, long-context research and multi-tool collaboration. This generation no longer supports the traditional expansion thinking method of manually setting budget_tokens, but instead uses adaptive thinking and effort parameters to control the depth of reasoning. It is suitable for building advanced agents that can disassemble tasks on their own, call tools, check results, and modify solutions. When migrating, you need to check the sampling parameters, structured output method, and maximum output length to avoid request failure caused by using the old API configuration.

Context

1000000 tokens

Max output

128000 tokens

Input

llm image

Output

llm