GPT-6 Astra hands-on: composes first, orchestrates later, and reportedly outshines Fable and Sol
paw_lean · x · 2026-09-14
A user tested the rumored GPT-6 "Astra" 3D capabilities by having it build a 3D orchestra performing a concerto it composed itself.
- The author found Astra's compositions clearly stronger than Fable and Sol: a huge improvement over Sol and in parts better than Fable
- The key difference is process: Astra starts with a full composition, then spends a separate pass refining voicing and accompaniment — a compose-then-refine workflow
- Unofficial hands-on/leak content; the model's authenticity is unverified
More from Models
- David Bellamy clarifies his experiment used K2 Horizon, an open-weights 375B LLM — JeremyNguyenPhD · 2026-09-14
- Swift-Qwen3.8-27b, a token-efficient reasoning Qwen finetune, trends on Hugging Face — ukisai · 2026-09-14
- Researcher posts proof he both synthesized viruses and trained a 375B open-weight LLM — ethanCaballero · 2026-09-14
- Toby Ord: 10x more RLVR compute cuts tokens-to-target ~3x; gains may be math-specific — tobyordoxford · 2026-09-14
- New scaling curve has half the slope: 10,000x compute for 20%-to-80%, but bigger generational jumps — tobyordoxford · 2026-09-14
- Fudan NLP paper explains why max reasoning settings can backfire on SWE benchmarks — karminski3 · 2026-09-14