Unverified50% confidenceBenchmarkExact time
Claude Opus 4.7中间检查点通过率达57%，最终准确率却只有39.8%
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/9/2026
First Seen
Valid until: 10/7/2026
Sources
腾讯DiscoBench：评测LLM搜索代理歧义澄清能力的首个基准
bilibili熊二等兵7/4/2026
Related Claims
UnverifiedClaude Opus 4.6在SWE-Bench Pro测试中得分为53.4%78% similarPartially VerifiedClaude Opus 4.8在SWE-bench基准测试中得分69.2%，GPT 5.5为58.6%，Google Gemini为54.2%76% similarUnverified据非官方爆料，Terminal Bench 3.0基准测试中Claude Opus 5 Max以43.5%的得分超越GPT-5.6-Sol的34.6%重登榜首73% similarUnverifiedClaude Opus 4.7在SWE Bench Pro上得分从53.4%提升至54.3%73% similarUnverified在同一模型（Claude 4.6 Opus）下，OCR的精度（Precision）为33.9%，通用Agent为7.23%，OCR精度提升约4.7倍72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/387224API
curl https://kongchang.com/api/v1/knowledge/claims/387224MCP
get_claim(id=387224)