gemini-4IMAGE
Gemini 4 Argon 跑分碾压 GPT-6 Astra
列出 Argon 的 DeepSWE 77.9%、Vals Index 第一、AutomationBench 51.3%、LVBench 91.7% 等成绩,称其反超 GPT-6 Astra 与 Opus 5.5。
Salio
@Mr_Salio28 天前收录原推发布于 2026.10.01

Salio @Mr_Salio原推发布于 2026.10.01
🚨 Gemini 4 argon beacmarks are terrifying It's smashed Gpt-6 astra and fable 5.1 and Opus 5.5 🤯 • 77.9% DeepSWE v1.1, new SOTA • #1 on Vals Index • 51.3% AutomationBench, #1 • 91.7% LVBench, new SOTA • 68% CWE-bench v1, tied #1 What unexpected comeback Holy shit
这条是引用着另一条写的:x.com/mr_salio/status/2105390488983347209
x.com/mr_salio/status/2105391421129625969采集于 2026.10.01 10:01