Side-by-side comparison of AI model specs, benchmarks, and pricing.
| Spec | Mistral Large 3 Mistral | Gemini 3.1 Pro Google |
|---|---|---|
| Tier | open | frontier |
| Release Date | 2025-12-02 | 2026-02-19 |
| Context Window | 256K | 1049K |
| Max Output | 16K | 66K |
| Input Price (/1M) | $0.5 | $2 |
| Output Price (/1M) | $1.5 | $12 |
| Arena Elo | 1,414 | 1,500 |
| MMLU | 84% | 92.6% |
| GPQA | 43.9% | 94.3% |
| MATH | — | 96.8% |
| HumanEval | 92% | 94.6% |
| SWE-bench | — | 80.6% |
| AIME | 42% | 91.2% |
| SimpleQA | 29% | 79.6% |
| Capabilities | vision, tool-use, code | vision, tool-use, code, reasoning, agentic, audio |