Up to four models
GPT-6.1 Sol vs Gemini 3.8 Flash
On the index (GPT-4 = 100), GPT-6.1 Sol scores 196 and Gemini 3.8 Flash 175. 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
Gemini 3.8 Flash175159–191
WeightsClosed
ReleasedSep 2026
Price in / out$0.82 / $4.13
Context1.05M
Core tests taken7 of 9
GPT-6.1 Sol leads Gemini 3.8 Flash by 21 points, ahead on 3 of 6 shared core tests, at 2.4× the price.
Test by testthird-party results
GPT-6.1 SolGemini 3.8 Flash
Gemini 3.8 FlashEvery result as a table
| Test | GPT-6.1 Sol | Gemini 3.8 Flash |
|---|---|---|
| Reasoning & knowledge | ||
| GPQA Diamond (Epoch AI) | 95.4% | 95.4% |
| SimpleQA Verified | 73.9% | 69.7% |
| Maths | ||
| FrontierMath (tiers 1-3) | 93.7% | 68.4% |
| AIME-style maths (OTIS mock) | 100.0% | 98.9% |
| ProofBench v1.1 | 99.0% | 48.0% |
| Chess puzzles | 68.0% | 58.0% |
| FrontierMath (tier 4) | 100.0% | 21.9% |
| Coding | ||
| Terminal-Bench 4.0 | 55.0% | 19.7% |
| Code Migration | 65.1% | 36.5% |
| CyberBench v1.1 | 39.3% | 43.8% |
| IOI | 96.9% | 56.9% |
| ProgramBench | 3.0% | 1.0% |
| Vibe Code Bench v1.1 | 88.9% | 78.6% |
| Agents | ||
| APEX-Agents | 60.0% | 64.3% |
| BioMysteryBench | 79.6% | 62.2% |
| SRE Bench | 50.8% | 11.1% |
| Terminal-Bench-Science | 52.9% | 8.6% |
| DTBench | 96.5% | 95.7% |
| Furniture assembly | 80.0% | 31.7% |
| LMCA | 62.8% | 52.9% |
| Mystery game puzzles | 80.0% | 47.0% |
| Professional work | ||
| Finance Agent (v2) | 52.0% | 61.4% |
| EMB | 70.8% | 72.2% |
| HLAB-Vals-1.1 | 6.3% | 12.1% |
| Legal Research Bench | 38.5% | 38.9% |
| MedCode | 48.8% | 48.1% |
| MedScribe | 86.5% | 84.5% |
| Public Benefits Bench v1.1 | 59.3% | 65.3% |
| SAGE | 46.5% | 35.1% |
| Tax Agent Bench | 21.8% | 32.1% |
| Novel problems | ||
| ARC-AGI-2 | 94.2% | 89.2% |
| ARC-AGI-1 | 98.5% | 98.5% |
| ARC-AGI-3 | 52.7% | 10.4% |
| MysteryMechanism | 46.4% | 36.5% |
Tests at least two of these models have taken. Bold rows are the core tests.
