Side-by-side comparison of AI model specs, benchmarks, and pricing.
| Spec | Claude Opus 4.6 Anthropic | Gemini 3 Flash Google |
|---|---|---|
| Tier | frontier | frontier |
| Release Date | 2026-02-05 | 2025-12-17 |
| Context Window | 200K | 1049K |
| Max Output | 128K | 66K |
| Input Price (/1M) | $5 | $0.5 |
| Output Price (/1M) | $25 | $3 |
| Arena Elo | 1,496 | 1,470 |
| MMLU | 92.5% | 88% |
| GPQA | 91.3% | 80% |
| MATH | 97.6% | 90% |
| HumanEval | 97% | 94% |
| SWE-bench | 72.5% | 57% |
| AIME | 83.3% | 72% |
| SimpleQA | 43.2% | 36% |
| Capabilities | vision, tool-use, code, extended-thinking, agentic | vision, tool-use, code, reasoning, audio |