Unverified60% confidenceFactTime unknown
Claude 3.7 Sonnet的推理模式每次使用消耗两个请求额度
2
Sources
60%
Confidence
Medium-term (~90 days)
Relevance
7/2/2026
First Seen
Valid until: 9/30/2026
Sources
Related Claims
Unverified以 Claude 3.5 Sonnet 为代表的模型在代码生成的一次通过率（pass@1）上达到了工程实用水准69% similarUnverifiedClaude Sonnet 3.7以扩展思维（Extended Thinking）能力著称，在SWE-bench编程基准测试上取得了当时的最高分67% similarUnverified前沿思考型LLM（如Claude 3.5 Sonnet）能以较高一致性遵循约150-200条指令65% similarUnverifiedClaude 3.5 Sonnet在SWE-bench软件工程基准测试上突破49%的解题率65% similarUnverifiedClaude 3.5 Sonnet在工具调用的准确率和参数生成的可靠性方面表现突出65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/58838API
curl https://kongchang.com/api/v1/knowledge/claims/58838MCP
get_claim(id=58838)