Hands-on: fable still beats astra for daily tasks while sol medium shines
mertdumenci · x · 2026-09-06
User mertdumenci compares several video models in daily use: fable still outperforms astra on most everyday tasks, astra remains weak on UI taste, while sol medium is "phenomenal" for lighter tasks where fable isn't worth pulling out.
He describes current capabilities as spiky, with each model having distinct strengths.
More from Models
- Miles Brundage asks if GPT-6 is the public name and Astra the insider codename — Miles_Brundage · 2026-09-06
- Dev wowed as Astra nails a full Three.js run and shows off Blender skills — victormustar · 2026-09-06
- Clinical benchmarks show Astra is incremental, still trailing Anthropic's frontier — danielmckinn0n · 2026-09-06
- Zvi's framework: honeypot attempt reductions up to 50-75% are fine, beyond that alarming — TheZvi · 2026-09-06
- GPT-3 to GPT-6: OpenAI API prices dropped up to 6x in five years — BLUECOW009 · 2026-09-06
- Astra called OpenAI's worst release in a while: 'severely undercooked' at coding — tokenbender · 2026-09-06