WHAT NEW IN X11.03 TUE

X 上的 AI 合集档案 · 22 个合集 · 66,612 条

gpt-6IMAGE

Muse Spark 1.3 排名压过 GPT-6 Astra 引争议

有人质疑:Muse Spark 1.3 在智能指数上比 Astra 高一分,但 Astra 在 ARC-AGI-3、FrontierMath 上成绩惊人,说明实战测试比理论榜单更重要。

Veee
@vikktorrrre2026.09.17收录原推发布于 2026.09.16
Muse Spark 1.3 排名压过 GPT-6 Astra 引争议
Veee @vikktorrrre原推发布于 2026.09.16
在 X 上打开原推

How is Muse Spark 1.3 ranked above GPT-6 Astra, yet no one sees it as AGI potential If Jensen Huang claimed Astra is AGI, then a model ranked above it on the Artificial Analysis Intelligence Index should be interpreted as it's pro AGI ye? This just goes to show that practical tests should be valued above theoretical ones. It ranked: • Muse Spark 1.3: 62 • GPT-6 Astra: 61 Yet GPT-6 Astra got: • 98.6% on ARC-AGI-3 • 97.6% on FrontierMath • 100% on ExploitBench Trump doesn't even regard Meta's AI tool as revolutionary. And it's geo-blocked in some regions.

x.com/vikktorrrre/status/2100180857918980373采集于 2026.09.17 10:18