Side-by-side comparison of AI model specs, benchmarks, and pricing.
| Spec | Gemini 2.5 Pro Google | Claude Sonnet 4.6 Anthropic |
|---|---|---|
| Tier | frontier | frontier |
| Release Date | 2025-06-05 | 2026-02-17 |
| Context Window | 1049K | 200K |
| Max Output | 66K | 64K |
| Input Price (/1M) | $1.25 | $3 |
| Output Price (/1M) | $10 | $15 |
| Arena Elo | 1,450 | 1,470 |
| MMLU | 89% | 91% |
| GPQA | 74% | 88% |
| MATH | 90.2% | 97.8% |
| HumanEval | 93.2% | 96% |
| SWE-bench | 63.8% | 70.3% |
| Capabilities | vision, tool-use, code, reasoning, audio | vision, tool-use, code, extended-thinking |