Dev's hands-on take: Astra feels like a downgrade from Sol
mertdumenci · x · 2026-09-12
A developer shares a blunt hands-on take on Google's new model: Astra feels like a downgrade from Sol, except when you can run 500 trillion instances of it to brute-force counterexamples in natural-language proofs. He adds that scrolling X makes him sad, but actually using the model frustrates him more—proof there's still a long way to go.
Related event: Developers Call New Model Astra a Downgrade from Sol(3 posts)→
More from Models
- tldraw founder lists 10 bugs in ChatGPT's new sketch feature — manosaie · 2026-09-12
- GPT-6 Astra tops DDD benchmark for multi-step retrosynthesis, nearing specialist models — CatAstro_Piyush · 2026-09-12
- User pleads for separate Codex credit limits as shared quota halts scheduled agents — koltregaskes · 2026-09-12
- OpenAI's GPT-Live-1 tested: ServiceTitan demos hands-free voice agent for rooftop technicians — pbbakkum · 2026-09-12
- Musk delays Grok 4.7 by days, says RL over-penalized response length — inductionheads · 2026-09-12
- DeepSeek Harness repo shows Desktop client code, fueling speculation — teortaxesTex · 2026-09-12