Up to four models
GPT-6.1 Sol vs Claude Haiku 5.5
On the index (GPT-4 = 100), GPT-6.1 Sol scores 196 and Claude Haiku 5.5 174. 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 Sol
196180–212
WeightsClosed
ReleasedSep 2026
Price in / out$2.0 / $10
Context1.05M
Core tests taken6 of 9
Claude Haiku 5.5
174158–191
WeightsClosed
ReleasedOct 2026
Price in / out$0.10 / $0.50
Context1M
Core tests taken7 of 9
GPT-6.1 Sol leads Claude Haiku 5.5 by 22 points, ahead on 5 of 6 shared core tests, at 20× the price.
Test by testthird-party results
GPT-6.1 SolClaude Haiku 5.5
Every result as a table
| Test | GPT-6.1 Sol | Claude Haiku 5.5 |
|---|---|---|
| Reasoning & knowledge | ||
| GPQA Diamond (Epoch AI) | 95.4% | 89.6% |
| LiveBench Data Analysis | 82.7% | 72.8% |
| LiveBench IF | 74.2% | 66.5% |
| LiveBench Language | 90.1% | 63.5% |
| LiveBench Reasoning | 92.6% | 84.6% |
| EBR-bench | 54.3% | 43.8% |
| SimpleQA Verified | 73.9% | 23.8% |
| Maths | ||
| FrontierMath (tiers 1-3) | 93.7% | 75.1% |
| AIME-style maths (OTIS mock) | 100.0% | 98.9% |
| LiveBench Mathematics | 96.8% | 92.8% |
| ProofBench v1.1 | 99.0% | 89.0% |
| Chess puzzles | 68.0% | 55.0% |
| FrontierMath (tier 4) | 100.0% | 46.3% |
| Coding | ||
| Terminal-Bench 4.0 | 55.0% | 36.4% |
| LiveBench Agentic Coding | 56.8% | 57.2% |
| LiveBench Coding | 80.7% | 77.9% |
| Code Migration | 65.1% | 44.6% |
| CyberBench v1.1 | 39.3% | 75.5% |
| IOI | 96.9% | 47.3% |
| ProgramBench | 3.0% | 2.5% |
| Vibe Code Bench v1.1 | 88.9% | 90.4% |
| Agents | ||
| APEX-Agents | 60.0% | 56.1% |
| BioMysteryBench | 79.6% | 70.4% |
| SRE Bench | 50.8% | 5.7% |
| Terminal-Bench-Science | 52.9% | 10.0% |
| DTBench | 96.5% | 95.0% |
| Furniture assembly | 80.0% | 47.5% |
| LMCA | 62.8% | 51.1% |
| Mystery game puzzles | 80.0% | 30.0% |
| Professional work | ||
| Finance Agent (v2) | 52.0% | 54.1% |
| EMB | 70.8% | 59.0% |
| HLAB-Vals-1.1 | 6.3% | 4.2% |
| Legal Research Bench | 38.5% | 43.8% |
| MedCode | 48.8% | 51.5% |
| MedScribe | 86.5% | 89.5% |
| Public Benefits Bench v1.1 | 59.3% | 63.2% |
| SAGE | 46.5% | 45.8% |
| Tax Agent Bench | 21.8% | 31.9% |
| Novel problems | ||
| ARC-AGI-2 | 94.2% | 61.3% |
| ARC-AGI-1 | 98.5% | 89.0% |
| ARC-AGI-3 | 52.7% | 0.5% |
| MysteryMechanism | 46.4% | 35.1% |
Tests at least two of these models have taken. Bold rows are the core tests.
