Fable 5.1 draws early buzz: Astra makes 3.2× more agentic calls on grunt-work task
repligate · x · 2026-09-05
repligate says he expects to love Fable 5.1 despite not having tried it yet. The quoted benchmark: on a self-portrait task requiring sustained grunt work, Astra made 642 tool calls vs 204 for Fable 5.1 (about 3.2×) — Astra pushes through tasks with less nudging.
More from Models
- GPT-6 Pro Spotted Available in ChatGPT, Developer Claims — johnlindquist · 2026-09-05
- ProgramBench metric split: DeepSeek and GPT shine on raw pass rate, Chinese models lag — teortaxesTex · 2026-09-05
- GPT-6 Astra leak demo shows impressive UI generation capabilities — oscardumlao · 2026-09-05
- Plus user burns through 5-hour limit in just 2 prompts, down to 2% on medium thinking — studiocookies_ · 2026-09-05
- GPT-6 Astra burns 80% of weekly usage limit in just two hours, user reports — BLUECOW009 · 2026-09-05
- Same prompt, different personality: users say the new model is a downgrade for writing — WarComfortable2502 · 2026-09-05