Up to four models
Claude Fable 5.2 vs Claude Opus 5.5
On the index (GPT-4 = 100), Claude Fable 5.2 scores 202 and Claude Opus 5.5 195. 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 Opus 5.5
195178–211
WeightsClosed
ReleasedSep 2026
Price in / out$4.0 / $20
Context1M
Core tests taken6 of 9
Claude Fable 5.2 and Claude Opus 5.5 are too close to call (7 points apart, within the uncertainty), ahead on 3 of 4 shared core tests, at 2.0× the price.
Test by testthird-party results
Claude Fable 5.2Claude Opus 5.5
Every result as a table
| Test | Claude Fable 5.2 | Claude Opus 5.5 |
|---|---|---|
| Reasoning & knowledge | ||
| GPQA Diamond (Epoch AI) | 96.0% | 94.1% |
| EBR-bench | 90.5% | 71.4% |
| SimpleBench | 94.9% | 88.4% |
| SimpleQA Verified | 94.0% | 72.2% |
| Maths | ||
| FrontierMath (tiers 1-3) | 95.1% | 91.2% |
| AIME-style maths (OTIS mock) | 99.7% | 100.0% |
| ProofBench v1.1 | 100.0% | 100.0% |
| Chess puzzles | 79.0% | 67.0% |
| FrontierMath (tier 4) | 100.0% | 95.0% |
| Coding | ||
| Terminal-Bench 4.0 | 65.1% | 65.1% |
| ProgramBench | 19.0% | 18.5% |
| Vibe Code Bench v1.1 | 93.8% | 90.3% |
| FrontierSWE | 69.9% | 62.3% |
| Agents | ||
| APEX-Agents | 80.1% | 73.5% |
| BioMysteryBench | 83.7% | 79.3% |
| Terminal-Bench-Science | 55.7% | 47.1% |
| DTBench | 99.2% | 98.9% |
| Furniture assembly | 90.0% | 83.3% |
| LMCA | 70.6% | 68.2% |
| Mystery game puzzles | 90.0% | 71.0% |
| Professional work | ||
| HLAB-Vals-1.1 | 5.0% | 9.2% |
| Legal Research Bench | 46.2% | 50.5% |
| Tax Agent Bench | 32.1% | 44.5% |
| Novel problems | ||
| MysteryMechanism | 51.8% | 49.5% |
Tests at least two of these models have taken. Bold rows are the core tests.