Unverified75% confidenceFactTime unknown
DeepSeek与Claude Sonnet在SWE-bench编码基准测试上的差距仅在1%以内
1
Sources
75%
Confidence
Medium-term (~90 days)
Relevance
5/31/2026
First Seen
Valid until: 8/29/2026
Sources
DeepSeek-TUI：免费Claude Code替代品，成本低20倍登顶GitHub
bilibili侧光译室
Related Entities
Related Claims
UnverifiedClaude Sonnet 4.6与GPT-5.1 Codex在SWE-Bench Verified上的解决率差距仅为3.3个百分点67% similarUnverifiedDeepSeek-V3发布时在多项基准测试中与Claude 3.5 Sonnet持平甚至超越，训练成本据报道仅为主流闭源模型的数分之一67% similarUnverifiedClaude Code + DeepSeek稳定性测试得分100分，Codex + DeepSeek仅得40分67% similarUnverifiedGLM-4.7在代码能力基准测试中超过DeepSeek V3.2及Claude Sonnet 4.562% similarUnverifiedSonnet 5在CursorBench编程基准上仅排在第13位61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/15204API
curl https://kongchang.com/api/v1/knowledge/claims/15204MCP
get_claim(id=15204)