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-oss-20b
| Claude Haiku 4.5 Anthropic | gpt-oss-20b OpenAI | |
|---|---|---|
| Pricing | ||
| Input $/1M | $1.00 | $0.03 |
| Output $/1M | $5.00 | $0.13 |
| Blended $/1M (3:1) | $2.00 | $0.055 |
| Cache read $/1M | $0.10 | $0.03 |
| Cache write $/1M | $1.25 | — |
| Quality indices | ||
| Intelligence Index | 16.9 | 9 |
| Coding Index | 43.9 | 20.7 |
| Math Index | — | 89.3 |
| Agentic Index | 8 | 1.2 |
| Speed & latency | ||
| Output speed (tok/s) | — | 189 |
| Time to first token | — | 0.47s |
| Time to first answer token | — | 11.04s |
| Benchmarks (Artificial Analysis) | ||
| MMLU-Pro | — | 74.8% |
| GPQA Diamond | — | 68.8% |
| Humanity's Last Exam | — | 11.0% |
| LiveCodeBench | — | 77.7% |
| SciCode | — | 38.9% |
| AIME 2025 | — | 89.3% |
| IFBench | — | 65.1% |
| AA-LCR (Long Context Reasoning) | — | 34.7% |
| Terminal-Bench Hard | — | 10.6% |
| Terminal-Bench 2.1 | — | 13.9% |
| τ²-Bench (Telecom) | — | 60.2% |
| τ³-Bench Banking | — | 7.0% |
| Design Arena (Elo) | ||
| Website | 1,135 | 865 |
| UI components | 1,115 | — |
| Data viz | 1,139 | 946 |
| SVG | 1,049 | — |
| 3D | 1,101 | — |
| Game dev | 1,120 | — |
| Code categories | 1,130 | — |
| ASCII art | 1,159 | — |
| Specs | ||
| Context window | 200K | 131K |
| Max output tokens | 64K | 118K |
| Input modalities | text, image, file | text |
| Tokenizer | Claude | GPT |