Atlas

Benchmarks

← All benchmarks

RuneBench Gold - Smith + Alch 15m

Games · 2026-02-26

RuneBench Smith + Alch 15m gold-accumulation task. Agents start from a smithing and alchemy-oriented condition and are scored by peak coins seen during the run.

Top models (higher is better)

ModelScore
Opus 4.5452
Haiku 4.5234
Opus 4.7195
Sonnet 4.681.0
Sonnet 4.521.0
GPT-5.4 Nano0.0
Loading Atlas data…