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 V3.1 TerminusGPT-5.6 Sol
| DeepSeek V3.1 Terminus DeepSeek | GPT-5.6 Sol OpenAI | |
|---|---|---|
| Pricing | ||
| Input $/1M | $0.27 | $5.00 |
| Output $/1M | $1.00 | $30.00 |
| Blended $/1M (3:1) | $0.453 | $11.25 |
| Cache read $/1M | $0.135 | $0.50 |
| Cache write $/1M | — | $6.25 |
| Quality indices | ||
| Intelligence Index | 21.4 | 58.9 |
| Coding Index | 43.5 | 77.4 |
| Math Index | 53.7 | — |
| Agentic Index | 18.1 | 54 |
| Speed & latency | ||
| Output speed (tok/s) | 0 | 67 |
| Time to first token | 0s | 120.38s |
| Time to first answer token | 0s | 120.38s |
| Benchmarks (Artificial Analysis) | ||
| MMLU-Pro | 83.6% | — |
| GPQA Diamond | 75.1% | 94.1% |
| Humanity's Last Exam | 8.4% | 47.2% |
| LiveCodeBench | 52.9% | — |
| SciCode | 32.1% | 56.1% |
| AIME 2025 | 53.7% | — |
| IFBench | 41.2% | 72.7% |
| AA-LCR (Long Context Reasoning) | 43.3% | 73.7% |
| Terminal-Bench Hard | 31.8% | 65.9% |
| Terminal-Bench 2.1 | — | 88.0% |
| τ²-Bench (Telecom) | 37.1% | 85.1% |
| τ³-Bench Banking | — | 33.0% |
| Design Arena (Elo) | ||
| Website | 1,214 | — |
| UI components | 1,224 | — |
| Data viz | 1,200 | — |
| SVG | 1,112 | — |
| 3D | 1,199 | — |
| Game dev | 1,189 | — |
| Code categories | 1,209 | — |
| Specs | ||
| Context window | 164K | 1.05M |
| Max output tokens | 33K | 128K |
| Input modalities | text | file, image, text |
| Tokenizer | DeepSeek | GPT |