opus-5.5VIDEO
Same prompt, different results: Opus 5.5 launch day vs today
On the exact same prompt, Opus 5.5's output today looks very different from launch day, with its NerfBench score falling from 100% on Sep 22 to 94.2% on Oct 2.
Md Ismail Šojal 🕷️
@0x0SojalSecArchived 2026.10.04Originally posted 2026.10.04
Md Ismail Šojal 🕷️ @0x0SojalSecOriginally posted 2026.10.04
Opus 5.5 launch vs today on the exact same prompt. It doesn’t look like the same model anymore. - drop on NerfBench. - Sep 22 launch: 100% - Oct 2: 94.2%
This post quotes another post: “Opus 5.5 drops on NerfBench as Astra holds 98%”
x.com/0x0sojalsec/status/2106437311076638739Archived 2026.10.04 10:03