Astra hits 74% on DeepSWE with 30k tokens, half the cost steps of GPT-5.6 Sol
haider1 · x · 2026-09-05
A comparison on the DeepSWE benchmark shows Astra scoring 74% using only 30k output tokens across 29 steps ($6.52), while GPT-5.6 Sol reaches a similar 73% but needs 60k tokens and 61 steps ($6.46) — roughly double the token and step count. The poster marvels at Astra's efficiency, especially given Sol launched just a few months ago.
More from Models
- antirez: Astra is a big jump for software dev as RLVR-scaled models keep growing — antirez · 2026-09-05
- Heavy user: ChatGPT's Astra intuitively outclasses Claude Fable 5.1 at the same price — DynaBeast · 2026-09-05
- Fable 5.1 refuses questions on Evoscale/Biohub papers, drawing 'Opus'ed' quip — nathanbenaich · 2026-09-05
- GPT-6 Astra and Claude give opposite answers to the same simple PC check — Jardani_xx · 2026-09-05
- xAI puts Grok 4.6 on its API with a claimed 500K-token context window — emmanuelvivier · 2026-09-05
- Anthropic unveils Enterprise Frontier Safeguards for Claude — emmanuelvivier · 2026-09-05