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.
DeepSeek V4 ProGPT-5.6 Sol
| DeepSeek V4 Pro DeepSeek | GPT-5.6 Sol OpenAI | |
|---|---|---|
| Pricing | ||
| Input $/1M | $0.435 | $5.00 |
| Output $/1M | $0.87 | $30.00 |
| Blended $/1M (3:1) | $0.544 | $11.25 |
| Cache read $/1M | $0.0036 | $0.50 |
| Cache write $/1M | — | $6.25 |
| Quality indices | ||
| Intelligence Index | 44.3 | 58.9 |
| Coding Index | 59.4 | 77.4 |
| Math Index | — | — |
| Agentic Index | 36.4 | 54 |
| Speed & latency | ||
| Output speed (tok/s) | 74 | 67 |
| Time to first token | 0.99s | 120.38s |
| Time to first answer token | 60.41s | 120.38s |
| Benchmarks (Artificial Analysis) | ||
| GPQA Diamond | 88.8% | 94.1% |
| Humanity's Last Exam | 35.9% | 47.2% |
| SciCode | 50.0% | 56.1% |
| IFBench | 76.5% | 72.7% |
| AA-LCR (Long Context Reasoning) | 66.3% | 73.7% |
| Terminal-Bench Hard | 46.2% | 65.9% |
| Terminal-Bench 2.1 | 64.0% | 88.0% |
| τ²-Bench (Telecom) | 96.2% | 85.1% |
| τ³-Bench Banking | 25.8% | 33.0% |
| Design Arena (Elo) | ||
| Website | 1,261 | — |
| Web apps | 1,005 | — |
| Full-stack | 948 | — |
| UI components | 1,263 | — |
| Data viz | 1,212 | — |
| SVG | 1,183 | — |
| 3D | 1,310 | — |
| Game dev | 1,288 | — |
| Godot game dev | 1,059 | — |
| Code categories | 1,272 | — |
| ASCII art | 1,194 | — |
| Specs | ||
| Context window | 1.05M | 1.05M |
| Max output tokens | 384K | 128K |
| Input modalities | text | file, image, text |
| Tokenizer | DeepSeek | GPT |