Up to four models
GPT-6.1 Terra vs Claude Fable 5
On the index (GPT-4 = 100), GPT-6.1 Terra scores 186 and Claude Fable 5 183. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.
GPT-6.1 Terra
186170–203
WeightsClosed
ReleasedOct 2026
Price in / out$1.0 / $5.0
Context1.05M
Core tests taken4 of 9
Claude Fable 5
183167–199
WeightsClosed
ReleasedJun 2026
Price in / out$10 / $50
Context1M
Core tests taken6 of 9
GPT-6.1 Terra and Claude Fable 5 are too close to call (3 points apart, within the uncertainty), ahead on 4 of 4 shared core tests, and costs 10× less.
Test by testthird-party results
GPT-6.1 TerraClaude Fable 5
Every result as a table
| Test | GPT-6.1 Terra | Claude Fable 5 |
|---|---|---|
| Reasoning & knowledge | ||
| GPQA Diamond (Epoch AI) | 94.3% | 93.9% |
| EBR-bench | 78.1% | 39.5% |
| SimpleBench | 72.2% | 81.9% |
| SimpleQA Verified | 57.3% | 70.7% |
| Maths | ||
| FrontierMath (tiers 1-3) | 90.9% | 87.0% |
| AIME-style maths (OTIS mock) | 100.0% | 100.0% |
| FrontierMath (tier 4) | 92.7% | 90.2% |
| Coding | ||
| Terminal-Bench 4.0 | 49.5% | 41.4% |
| Vibe Code Bench v1.1 | 84.0% | 90.3% |
| MirrorCode | 76.7% | 70.0% |
| Agents | ||
| APEX-Agents | 68.3% | 63.6% |
| Furniture assembly | 79.2% | 35.8% |
| Mystery game puzzles | 74.0% | 52.0% |
| Professional work | ||
| Legal Research Bench | 26.0% | 49.5% |
| Tax Agent Bench | 17.1% | 38.3% |
Tests at least two of these models have taken. Bold rows are the core tests.