gpt-6IMAGE
Critic says GPT-6 Astra writes code in silicon, not human
The author argues a good coding model should speak human rather than silicon even without a skill, and says Opus 5.5 beats GPT-6 Astra on that front.
Xiuyu Li
@sheriyuoArchived 2026.09.25Originally posted 2026.09.24

Xiuyu Li @sheriyuoOriginally posted 2026.09.24
A necessary condition for a good coding model: it should speak human, not silicon, even without a skill. At least Opus 5.5 beat GPT-6 Astra on that front.
This post quotes another post: “Code Arena WebDev: Claude Opus 5.5 overtakes GPT-6 Astra”
x.com/sheriyuo/status/2102990883931259093Archived 2026.09.25 10:10