Claude Fable 5.1

Zvi Mowshowitz's capabilities review observes that Claude Fable 5.1 ships with reduced cache prices, the option of zero data retention, and substantially more lenient classifiers than Fable 5. Early anecdotal signals suggest the jump from GPT-5.6 Sol to Astra may be larger than the jump from Fable 5 to Fable 5.1, poten The review reports qualitative capability improvements including better writing, reduced Claudisms (though some persist), greater willingness to admit mistakes, simpler code outputs, less obnoxious safety classifiers, and a proactive tendency to find useful work when given high effort levels. Zvi's own testing found bo

Emergent.sh confirms Anthropic launched Claude Fable 5.1 on September 1, 2026, ending weeks of leak-driven speculation, with the model going straight to general availability rather than phased enterprise preview. Developers can call claude-fable-5-1 on the Claude API and through Amazon Bedrock, Google Cloud, and Micros The article documents Fable 5.1's pricing keeping Fable 5's headline rates at $10/M input and $50/M output tokens, with the consequential change being the cache-read rate cut 75% to $0.25 per million tokens (down from $1). It notes Fable 5.1 defaults to different effort levels depending on surface: High in Claude Code,

An analysis of Claude Fable 5.1's September 1, 2026 benchmark results finds the gains concentrate in long, multi-step agentic and scientific work rather than shorter tasks Fable 5 already handled well. On agentic science, Terminal-Bench-Science rises from 24.7% to 52.6%, and on business automation, AutomationBench jump Independent evaluators broadly corroborate Anthropic's direction: Artificial Analysis ranks Fable 5.1 first on its Intelligence Index at score 66, ARC-AGI places it at 97.5% on ARC-AGI-1, and Vals AI ranks it first on the Vals Index. The article also pushes back on a widely circulated 95% SWE-bench Verified figure, cla

Anthropic announced Claude Fable 5.1 and Claude Mythos 5.1 on September 1, 2026, positioning them as advanced models for coding and knowledge work with a June 2026 knowledge cutoff. Both share the same infrastructure but differ in safety guardrails: Fable 5.1 is available to individual and enterprise users, while Mytho On benchmarks, Fable 5.1 scored 52.6% on Terminal-Bench-Science 0.1 (more than double Fable 5, exceeding Opus 5's 29.0% and GPT-5.6 Sol's 22.4%) and 55.8% on Terminal-Bench 4.0. Cache read fees dropped 75%, lowering overall workload costs by up to 45%. Anthropic also introduced Enterprise Frontier Safeguards for enhanc

Coursiv's blog confirms Anthropic released Claude Fable 5.1 and Claude Mythos 5.1 on September 1, 2026, describing Fable 5.1 as the generally available version with production safeguards and Mythos 5.1 as the same underlying model with looser safeguards for verified cybersecurity defenders and life scientists. The post The article provides detailed benchmark deltas showing Fable 5.1 reaching 52.6% on Terminal-Bench-Science 0.1 (vs 24.7% for Fable 5), 55.8% on Terminal-Bench 4.0, 1853 on GDPval-AA v2, 31.4% on AutomationBench, 73.4% on CursorBench 3.2.0, 41.7% on OSWorld 2.0 strict, and 65.0% on Humanity's Last Exam with tools, all ev

Jake Handy's Substack "Model Drop" piece confirms Claude Fable 5.1 launched on September 1, 2026 alongside its twin Claude Mythos 5.1, with same weights, two sets of safeguards, and identical pricing to Fable 5 at $10/$50 per million input/output tokens. The article highlights a 75% cache-read price cut to $0.25 per mi The post details Fable 5.1's headline benchmark scores and deltas versus Fable 5, Opus 5, and GPT-5.6 Sol, including Terminal-Bench-Science 0.1 at 52.6% (+27.9 over Fable 5), Terminal-Bench 4.0 at 60.9% for Mythos 5.1 and 55.8% for Fable 5.1, and GDPval-AA v2 at 1853 (+29 over Opus). It documents Mythos 5.1's restricte

R&D World reports that Anthropic released Claude Fable 5.1 and Claude Mythos 5.1 on September 1, with Fable 5.1 scoring 52.6% on Terminal-Bench-Science 0.1, an agentic scientific research benchmark, against 24.7% for Fable 5, 29.0% for Claude Opus 5, and 22.4% for GPT-5.6 Sol, with Anthropic reporting a standard error Independent evaluation from Artificial Analysis ranked Fable 5.1 atop its Intelligence Index, but found that at maximum effort the model used roughly 1.7 times as many output tokens and cost 20% more per task than its predecessor, despite Anthropic's 75% cache-read price reduction coinciding with launch. Artificial Ana

The alphacorp.ai launch guide confirms Claude Fable 5.1 launched September 1, 2026 at Fable 5's $10/$50 per-million-token prices, with cache reads cut 75% to $0.25 per million tokens and a June 2026 knowledge cutoff—the most recent in Anthropic's current lineup. The model retains a 1M-token context window at standard p The guide details three breaking changes for existing Fable 5 code: forced tool use (tool choice 'any' or a named tool) now returns a 400 error; earlier Claude models cannot read Fable 5.1's thinking blocks (though Fable 5.1 can read theirs); and editing earlier conversation context now triggers stricter model-binding