Up to four models
Claude Fable 5 vs GLM-5.2
On the index (GPT-4 = 100), Claude Fable 5 scores 183 and GLM-5.2 154. 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
183167–199
WeightsClosed
ReleasedJun 2026
Price in / out$10 / $50
Context1M
Core tests taken6 of 9
154138–169
WeightsOpen
ReleasedJun 2026
Price in / out$0.0070 / $6.8
Context1.05M
Core tests taken6 of 9
Claude Fable 5 leads GLM-5.2 by 30 points, ahead on 5 of 5 shared core tests, at 12× the price.
Test by testthird-party results
Claude Fable 5
GLM-5.2
Every result as a table
| Test | Claude Fable 5 | GLM-5.2 |
|---|---|---|
| Reasoning & knowledge | ||
| GPQA Diamond (Epoch AI) | 93.9% | 91.9% |
| LiveBench Data Analysis | 80.5% | 73.7% |
| LiveBench IF | 75.8% | 62.3% |
| LiveBench Language | 90.7% | 76.2% |
| LiveBench Reasoning | 89.6% | 78.6% |
| GPQA Diamond (Vals AI) | 93.2% | 85.6% |
| MMLU Pro | 91.5% | 86.7% |
| EBR-bench | 39.5% | 9.5% |
| SimpleBench | 81.9% | 58.8% |
| SimpleQA Verified | 70.7% | 34.2% |
| Maths | ||
| FrontierMath (tiers 1-3) | 87.0% | 59.2% |
| AIME-style maths (OTIS mock) | 100.0% | 86.4% |
| LiveBench Mathematics | 96.0% | 89.8% |
| Chess puzzles | 47.0% | 22.0% |
| FrontierMath (tier 4) | 90.2% | 29.3% |
| Surface Evolver Bench | 95.0% | 55.6% |
| Coding | ||
| LiveBench Agentic Coding | 62.2% | 51.8% |
| LiveBench Coding | 86.0% | 79.7% |
| Code Migration | 55.1% | 37.9% |
| LiveCodeBench | 89.8% | 69.5% |
| ProgramBench | 2.0% | 0.5% |
| SWE-bench | 95.0% | 82.8% |
| Vibe Code Bench v1.1 | 90.3% | 64.0% |
| DeepSWE | 69.9% | 43.8% |
| MirrorCode | 70.0% | 36.7% |
| WeirdML | 91.9% | 70.1% |
| Agents | ||
| APEX-Agents | 63.6% | 45.2% |
| DTBench | 98.4% | 93.6% |
| LMCA | 61.2% | 45.8% |
| Mystery game puzzles | 52.0% | 19.0% |
| Professional work | ||
| Finance Agent (v2) | 56.3% | 49.7% |
| CorpFin v2 | 71.8% | 66.1% |
| EMB | 73.7% | 61.5% |
| LegalBench | 88.6% | 84.1% |
| Legal Research Bench | 49.5% | 31.3% |
| MedCode | 56.1% | 40.8% |
| MedScribe | 88.5% | 83.5% |
| TaxEval v2 | 76.9% | 73.3% |
| Novel problems | ||
| ARC-AGI-2 | 89.2% | 22.8% |
| ARC-AGI-1 | 98.5% | 77.0% |
Tests at least two of these models have taken. Bold rows are the core tests.