Up to four models
Claude Fable 5.2 vs Claude Fable 5.1
On the index (GPT-4 = 100), Claude Fable 5.2 scores 202 and Claude Fable 5.1 189. 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 5.2
202185–218
WeightsClosed
ReleasedNov 2026
Price in / out$8.0 / $40
Context1M
Core tests taken4 of 9
Claude Fable 5.1
189173–205
WeightsClosed
ReleasedSep 2026
Price in / out$10 / $50
Context1M
Core tests taken6 of 9
Claude Fable 5.2 leads Claude Fable 5.1 by 13 points, ahead on 3 of 3 shared core tests, at a similar price.
Test by testthird-party results
Claude Fable 5.2Claude Fable 5.1
Every result as a table
| Test | Claude Fable 5.2 | Claude Fable 5.1 |
|---|---|---|
| Reasoning & knowledge | ||
| EBR-bench | 90.5% | 57.1% |
| SimpleBench | 94.9% | 86.6% |
| SimpleQA Verified | 94.0% | 70.8% |
| Maths | ||
| FrontierMath (tiers 1-3) | 95.1% | 90.2% |
| AIME-style maths (OTIS mock) | 99.7% | 100.0% |
| ProofBench v1.1 | 100.0% | 100.0% |
| Chess puzzles | 79.0% | 48.0% |
| FrontierMath (tier 4) | 100.0% | 87.8% |
| Coding | ||
| Terminal-Bench 4.0 | 65.1% | 58.1% |
| ProgramBench | 19.0% | 7.0% |
| Vibe Code Bench v1.1 | 93.8% | 90.3% |
| FrontierSWE | 69.9% | 56.3% |
| Agents | ||
| APEX-Agents | 80.1% | 68.6% |
| Terminal-Bench-Science | 55.7% | 40.0% |
| DTBench | 99.2% | 97.6% |
| Furniture assembly | 90.0% | 70.0% |
| LMCA | 70.6% | 65.5% |
| Mystery game puzzles | 90.0% | 58.0% |
| Professional work | ||
| Legal Research Bench | 46.2% | 55.3% |
| Tax Agent Bench | 32.1% | 50.7% |
| Novel problems | ||
| MysteryMechanism | 51.8% | 47.8% |
Tests at least two of these models have taken. Bold rows are the core tests.