Fable 5.1 release: Doubles scientific agent scores, cuts costs by 25%
mark_k · x · 2026-09-02
Anthropic released Fable 5.1, a major upgrade despite the minor version bump. It scores 52.6% on Terminal-Bench-Science (up from 24.7% for Fable 5) and shows significant gains on Terminal-Bench 4.0, CursorBench, and AutomationBench. It outperforms predecessors in coding, knowledge work, computer use, and reasoning. Costs are 25% lower for typical workloads and up to 45% lower for agentic workloads.
Related event: Anthropic Releases Fable 5.1 and Mythos 5.1 with 75% Cheaper Caching(174 posts)→
More from Models
- Mocked benchmark shows DeepSeek's document chart grounding badly misaligned — burkov · 2026-10-18
- Dev complains Claude's habit of saying "not done" is his biggest pain point — jasonkneen · 2026-10-18
- Open-weight TYGER speech decision model makes 80ms turn-taking calls for full-duplex voice agents — solyarisoftware · 2026-10-18
- Paying Google AI subscriber blocked: custom-domain Workspace accounts can't access AI Studio — ChrisRackauckas · 2026-10-18
- Running Ling 3.1 Flash (560B MoE, 25B Active) Across 4 DGX Sparks via llama.cpp RPC — BerryBlush_Rogers691 · 2026-10-18
- Users press OpenAI on long-delayed Adult Mode a year after it was promised — ___Patrice___ · 2026-10-18