Up to four models
Claude Fable 5.1 vs Claude Fable 5
On the index (GPT-4 = 100), Claude Fable 5.1 scores 192 and Claude Fable 5 186. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.
Claude Fable 5.1
192175–208
WeightsClosed
ReleasedSep 2026
Price in / out$10 / $50
Context1M
Core tests taken6 of 9
Claude Fable 5
186170–203
WeightsClosed
ReleasedJun 2026
Price in / out$10 / $50
Context1M
Core tests taken6 of 9
Claude Fable 5.1 leads Claude Fable 5 by 6 points, ahead on 5 of 5 shared core tests, at a similar price.
Test by testthird-party results
Claude Fable 5.1Claude Fable 5
Every result as a table
| Test | Claude Fable 5.1 | Claude Fable 5 |
|---|---|---|
| Reasoning & knowledge | ||
| LiveBench Data Analysis | 80.3% | 80.5% |
| LiveBench IF | 73.0% | 75.8% |
| LiveBench Language | 89.5% | 90.7% |
| LiveBench Reasoning | 91.7% | 89.6% |
| GPQA Diamond (Vals AI) | 93.4% | 93.2% |
| MMLU Pro | 92.4% | 91.5% |
| MMMU Pro | 90.6% | 89.3% |
| EBR-bench | 57.1% | 39.5% |
| SimpleQA Verified | 70.8% | 70.7% |
| Maths | ||
| FrontierMath (tiers 1-3) | 90.2% | 87.0% |
| AIME-style maths (OTIS mock) | 100.0% | 100.0% |
| LiveBench Mathematics | 97.0% | 96.0% |
| ProofBench v1.1 | 100.0% | 95.0% |
| Chess puzzles | 48.0% | 47.0% |
| FrontierMath (tier 4) | 87.8% | 90.2% |
| Coding | ||
| Terminal-Bench 4.0 | 58.1% | 41.4% |
| LiveBench Agentic Coding | 66.1% | 62.2% |
| LiveBench Coding | 86.4% | 86.0% |
| Code Migration | 54.6% | 55.1% |
| LiveCodeBench | 90.5% | 89.8% |
| ProgramBench | 7.0% | 2.0% |
| Vibe Code Bench v1.1 | 90.3% | 90.3% |
| FrontierSWE | 56.3% | 47.0% |
| GSO-Bench | 88.2% | 78.4% |
| WeirdML | 92.9% | 91.9% |
| Agents | ||
| APEX-Agents | 68.6% | 63.6% |
| Terminal-Bench-Science | 40.0% | 15.7% |
| DTBench | 97.6% | 98.4% |
| Furniture assembly | 70.0% | 35.8% |
| LMCA | 65.5% | 61.2% |
| Mystery game puzzles | 58.0% | 52.0% |
| Professional work | ||
| Finance Agent (v2) | 58.9% | 56.3% |
| EMB | 76.7% | 73.7% |
| LegalBench | 88.5% | 88.6% |
| Legal Research Bench | 55.3% | 49.5% |
| MedCode | 53.5% | 56.1% |
| MedScribe | 91.3% | 88.5% |
| MortgageTax | 70.8% | 68.9% |
| Public Benefits Bench v1.1 | 74.9% | 70.4% |
| SAGE | 48.5% | 51.9% |
| Tax Agent Bench | 49.2% | 38.3% |
| TaxEval v2 | 76.0% | 76.9% |
| Novel problems | ||
| ARC-AGI-2 | 90.0% | 89.2% |
| ARC-AGI-1 | 97.5% | 98.5% |
Tests at least two of these models have taken. Bold rows are the core tests.
