外部精选
软件工程
#开发者实时流Show HN: I found a more accurate upcoming LLM Benchmarks leaderboard
Share of 80 questions answered correctly · higher is better
Hacker News 最新techappsdev2026/10/27 · 今天1 分钟阅读
以下正文同步自 Hacker News 最新，版权归原站所有，已转换为易读排版。
Share of 80 questions answered correctly · higher is better
1Claude Opus 5.585.0
2Gemini 4 Argon84.8
3GPT-5.3-Codex82.5
3Gemini 3.8 Flash82.5
5Claude Fable 5.181.2
6Grok 4.780.0
7Muse Wonder78.8
7Claude Sonnet 5.578.8
9Gemini 3.1 Pro Preview77.5
9Muse Spark Open77.5
9Kimi K3.1 Pro77.5
12GPT-6 Astra76.2
12GPT-6.1 Terra Pro76.2
12GPT-6.1 Sol76.2
15GPT-5.6 Terra75.0
57 more models · see the full leaderboard
正文由 探How 从来源站点 RSS 同步，内容未经改写；遇到排版缺失或需要图片、视频时请以原文为准。