GPT-6 Sol vs MiniMax M3.1 Flash
GPT-6 Sol leads by 3.9 capability points (≈95% interval +1.5 to +6.4): a clear lead over MiniMax M3.1 Flash · P(higher) ≈ 100%
Wins 14 of 21 shared benchmarks (7 lost)
Each line runs through one model’s priced reasoning settings, cheapest to most expensive; grey points are other models’ priced settings in the same range. Results whose reasoning effort the source didn’t report are left out wherever a model has named effort levels. Capability and relative cost are the published estimates (relative cost is cost per task as a multiple of GPT-5 (high)); hover a setting, or use the arrow keys, to set it beside the other model’s same setting.
GPT-6 SolvsMiniMax M3.1 Flashcapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev
Benchmark wins14 – 7 · 21 shared
GPT-6 SolvsMiniMax M3.1 Flashwins by categoryData Nov 15, 2026Data: Mercor, Vals AI, Neel Nanda and 10 othersVillage LLM Atlasatlas.village.dev
Margins
Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is MiniMax M3.1 Flash minus GPT-6 Sol.
| ALE-Bench | Code | 2637 | 1495 | −40.4 |
| NoCoT-Bench NCKI (kspine v3) | General QA | 119 | 65.9 | −36.6 |
| Toloka Arena (v3) | Agents | 58.0 | 22.0 | −36.0 |
| BuseyBench SVG | Multimodal | 7.5 | 4.2 | −33.1 |
| GBENCH One-shot Coding GScore | Games | 70.1 | 41.1 | −28.9 |
| TaxCalcBench TY25 | Finance & Accounting | 28.0 | 0.0 | −28.0 |
| NoCoT-Bench NCRI 15.2 | General QA | 115 | 68.4 | −25.4 |
| Terminal-Bench 4.0 | Code | 44.4 | 25.3 | −19.2 |
| Vibe Code Bench v1.1 | Code | 87.8 | 73.9 | −13.9 |
| Tax Agent Bench | Finance & Accounting | 53.0 | 62.7 | +9.7 |
| LiveBench 2.0 | Agents | 81.9 | 73.1 | −8.8 |
| Harvey LAB (Mercor run, undisclosed ~213-task cohort) | Legal | 4.1 | 11.9 | +7.8 |
| τ³-Banking (Mercor run, bm25_grep retrieval) | Agents | 43.0 | 36.1 | −6.9 |
| Professional Reasoning Benchmark - Legal | Legal | 38.6 | 45.1 | +6.5 |
| Humanity's Last Exam Diamond | General QA | 33.8 | 28.1 | −5.7 |
| Legal Research Bench | Legal | 28.8 | 34.1 | +5.3 |
| Agents' Last Exam (ALE) | Agents | 31.6 | 27.0 | −4.6 |
| APEX-1 (Mercor, mean score) | Professional Work | 63.6 | 65.0 | +1.4 |
| Design Arena Overall Code Categories | Code | 1293 | 1297 | +0.7 |
| APEX-Agents 1.1 | Professional Work | 68.7 | 68.1 | −0.6 |
| APEX-Accounting 1.1 | Finance & Accounting | 59.7 | 60.0 | +0.3 |
GPT-6 SolvsMiniMax M3.1 Flashnormalised score marginsData Nov 15, 2026Data: Mercor, Vals AI, Neel Nanda and 10 othersVillage LLM Atlasatlas.village.dev