Anthropic
Claude Fable 5.1
Fable upgrade, about 25% cheaper for typical workloads. Anthropic's page says only “September 2026”; the day is from OpenRouter and Epoch AI.
- Announced
- Access
- Public
Also released as
- Claude Mythos 5.1 · Limited preview · 1 Sept 2026
Same model as Fable 5.1 with different safeguards. Only for trusted-access programmes in cybersecurity and life sciences.
Benchmark scores
| Test | Score | Best on the tracker | How it was measured |
|---|---|---|---|
| Epoch Capabilities Indexmany tests combined into one score | 167 points | GPT-6.2 Astra 174 points | Independentlisted as "Claude Fable 5.1", CI 164.29 to 170.2Run 9 Nov 2026 |
| AA Intelligence Index v5about ten tests in one score | 50 points | Claude Fable 5.2 64.6 points | Independentv5.0.1; listed as Claude Fable 5.1 (Xhigh, Default Fallback)Run 10 Nov 2026 |
| Remote Labor Indexpaid freelance projects | 18% | Claude Fable 5.2 41% | IndependentFable 5.1Run 17 Sept 2026 |
| FrontierSWE V2very long engineering jobs | 56% | Claude Fable 5.2 70% | Independentproximus harness, max effort, mean@5 over 34 tasks, 20-hour budget (best@5 66.7) |
| Terminal-Bench 4.1long command-line tasks | 58% | Claude Opus 5.5 66% | IndependentClaude Code harness, xhigh effort, 330 trials, ±3.31 (95% CI) |
| MirrorCoderebuilding whole programs | 73% | Claude Opus 5.5 79% | Independenthigh reasoning effort, MirrorCode (ML, +Private, 2L): 15 programs x 2 languages, 3 runs each, 10B-token / 7-day budget, ±9.2 SERun 10 Sept 2026 |
| APEX-Agentsprofessional office tasks | 69% | Gemini 4 Carbon 84% | Independentpass@1, unknown effort; listed as "Fable 5.1 [claude-fable-5.1]" |
| Humanity's Last Examexpert questions, no tools | 47% | GPT-6 Astra 55% | Independentleaderboard entry "Fable 5.1 (xhigh)" (claude-fable-5-1_xhigh) |
| FrontierMath T1–3hard maths problems | 90% | Claude Fable 5.2 95% | Independentmax reasoning effort (claude-fable-5-1_max)Run 1 Sept 2026 |
| FrontierMath T4research-level maths problemsSaturated | 88% | GPT-6.1 Sol 100% | Independentmax reasoning effort (claude-fable-5-1_max)Run 1 Sept 2026 |
| ARC-AGI-2visual puzzles, model aloneAverage person: 60%Saturated | 90% | GPT-6 Astra 95% | IndependentARC Prize config "Claude Fable 5.1 (XHigh)" (anthropic-claude-fable-5-1-xhigh) |
| OTIS Mock AIMEcompetition mathsSaturated | 100% | Top score | Independentmax reasoning effort (claude-fable-5-1_max)Run 1 Sept 2026 |
| METR time horizon (TH1.1 + TH2)task length an AI agent finishes half the time | 1.3 days | GPT-6 Astra 1.7 days | IndependentPooled TH1.1 + TH2 suites; 95% CI 718.7–18464.1 min |
Sources
- Introducing Claude Fable 5.1 and Claude Mythos 5.1Anthropic · Official · anthropic.com
- OpenRouter models API (created timestamp)OpenRouter · Dataset · openrouter.ai
- Epoch Capabilities IndexEpoch AI · Dataset · epoch.ai
- Epoch AI Benchmarking Hub (model release dates and runs)Epoch AI · Dataset · 9 Nov 2026 · epoch.ai
- Project Glasswing (Oct 6, 2026 Cyber Verification Program update)Anthropic · Official · anthropic.com