gpt-oss-20b beats Opus on a task while being 1,000x cheaper and faster

scaling01 · x · 2026-08-04

The author says they ran into a task where gpt-oss-20b outperformed Opus by a huge margin.

Their claim is blunt: the smaller model was not just better on this task, but also around 1,000x cheaper, faster, and more reliable.

Related event: gpt-oss-20b Outperforms Opus in Real-World Tasks at 1/1000th Cost(2 posts)→

Original post →

More from Models

Models channel →