Fable 5.1 hands-on: strongest coding model yet, using half the tokens of Opus 5
every · x · 2026-09-02
Every's hands-on review of Anthropic's Fable 5.1: more powerful than Fable 5, running for days on coding challenges that stumped its predecessor, while being fast and pleasant to talk to — a return to form after the odd tone of original Fable and Sonnet 5/Opus 5. It uses roughly half the tokens of Opus 5 for similar tasks, putting Fable-level work within an Opus budget. A new zero-data-retention option opens the door for enterprises. The authors call it the first taste of truly delegated AI for non-programmers — hand it a slide-deck task and come back in an hour or two. They also read the pre-Astra release as gamesmanship, hinting Fable 5.2 may be ready to go.
More from Models
- Fable 5.1 Beats May 2026 SOTA on Low-Thinking Image Benchmark — Afinetheorem · 2026-09-02
- OpenAI's Astra Model Imminent, Possibly Tomorrow — PathOfEnergySheild · 2026-09-02
- Running Fable 5.1 through AA's Intelligence Index cost $8,523 — 140M output tokens — ArtificialAnlys · 2026-09-02
- Claude Fable 5.1 leads GDPval-AA at 1853 Elo, edging Opus 5 within overlapping CIs — ArtificialAnlys · 2026-09-02
- Claude Fable 5.1's five effort levels span 11x in token usage, 58-66 on AA Index — ArtificialAnlys · 2026-09-02
- Fable 5.1 costs $3.76 per task; Anthropic's cache price cut saves $1.40 — ArtificialAnlys · 2026-09-02