Side-by-side comparison of AI model specs, benchmarks, and pricing.
| Spec | Gemini 2.5 Flash Google | GPT-5.3 Codex OpenAI |
|---|---|---|
| Tier | mid | frontier |
| Release Date | 2025-06-17 | 2026-02-05 |
| Context Window | 1049K | 400K |
| Max Output | 66K | 128K |
| Input Price (/1M) | $0.3 | $1.75 |
| Output Price (/1M) | $2.5 | $14 |
| Arena Elo | 1,410 | 1,480 |
| MMLU | 83.5% | 93% |
| GPQA | 65.8% | 81% |
| MATH | 82.1% | 96% |
| HumanEval | 90.3% | 93% |
| SWE-bench | — | 80% |
| AIME | 58% | 94% |
| SimpleQA | 28.3% | 58% |
| Capabilities | vision, tool-use, code, reasoning, audio | vision, tool-use, code, reasoning, agentic |