Compare models
Pick up to 4 models for a side-by-side spec sheet. Hover any dotted metric name for what the test measures. Best value in each row is highlighted.
Claude Haiku 4.5GPT-5.1
| Claude Haiku 4.5 Anthropic | GPT-5.1 OpenAI | |
|---|---|---|
| Pricing | ||
| Input $/1M | $1.00 | $1.25 |
| Output $/1M | $5.00 | $10.00 |
| Blended $/1M (3:1) | $2.00 | $3.44 |
| Cache read $/1M | $0.10 | $0.125 |
| Cache write $/1M | $1.25 | — |
| Quality indices | ||
| Intelligence Index | 29.6 | 36.9 |
| Coding Index | 43.9 | 49.4 |
| Math Index | — | 94 |
| Agentic Index | 16.4 | 21 |
| Speed & latency | ||
| Output speed (tok/s) | — | 110 |
| Time to first token | — | 41.75s |
| Time to first answer token | — | 41.75s |
| Benchmarks (Artificial Analysis) | ||
| MMLU-Pro | — | 87.0% |
| GPQA Diamond | — | 87.3% |
| Humanity's Last Exam | — | 26.5% |
| LiveCodeBench | — | 86.8% |
| SciCode | — | 43.3% |
| AIME 2025 | — | 94.0% |
| IFBench | — | 72.9% |
| AA-LCR (Long Context Reasoning) | — | 75.0% |
| Terminal-Bench Hard | — | 45.5% |
| Terminal-Bench 2.1 | — | 52.4% |
| τ²-Bench (Telecom) | — | 81.9% |
| τ³-Bench Banking | — | 14.0% |
| Design Arena (Elo) | ||
| Website | 1,149 | 1,215 |
| Web apps | — | 1,075 |
| Mobile apps | — | 1,122 |
| UI components | 1,142 | 1,209 |
| Data viz | 1,160 | 1,241 |
| SVG | 1,072 | 1,196 |
| 3D | 1,131 | 1,120 |
| Game dev | 1,155 | 1,237 |
| Code categories | 1,148 | 1,204 |
| ASCII art | 1,182 | 1,160 |
| Specs | ||
| Context window | 200K | 400K |
| Max output tokens | 64K | 128K |
| Input modalities | text, image, file | image, text, file |
| Tokenizer | Claude | GPT |