WHAT NEW IN X10.29 THU

X 上的 AI 合集档案 · 22 个合集 · 60,637 条

gpt-6IMAGE

GPT-6 Sol 修 bug 实测:分数低但便宜十倍

两个仓库 105 个隐藏 bug 测试中,GPT-6 Sol 得分 29.3 远低于 Astra 的 45,但 API 等效成本仅 9.93 美元,不到 Astra 的三分之一。

Paweł Huryn
@PawelHuryn2026.09.24收录原推发布于 2026.09.23
GPT-6 Sol 修 bug 实测:分数低但便宜十倍
Paweł Huryn @PawelHuryn原推发布于 2026.09.23
在 X 上打开原推

I finally tested GPT-6 Sol on a real work. 2 repos. 105 hidden bugs. Find and fixed what you can. It looks like a huge degradation. The results: - GPT-6 Astra (max): 45 - GPT-5.6 Sol (max): 43.5 - Opus 5.5 (max): 41.7 - Muse Spark 1.3 (max): 32.2 - GPT-6 Sol (max): 29.3 Until you measure the cost (API-equivalent): - GPT-6 Astra (max): $33.04 - GPT-5.6 Sol (max): $95.25 - Opus 5.5 (max): $58.53 - Muse Spark 1.3 (max): $18.11 - GPT-6 Sol (max): $9.93 More effort levels (xhigh, high, medium, low) dropping in this thread today 🧵

x.com/pawelhuryn/status/2102672181305958543采集于 2026.09.24 10:18