Tool Check
Best AI for math & reasoning
All tasks
Math & reasoning
Which model is best at math and hard problems?
Our pickGPT-6 AstraOpenAI · $10 in · $50 outThe best model anyone can use, with enough evidence. When several tie, the cheapest of them.
Best you can downloadDeepSeek V4 Pro 0813DeepSeek · $1.32 in · $3.96 out · downloadOpen weights: you can run it on your own computers.
Cheaper choiceGPT-6.1 SolOpenAI · $2 in · $10 outIn the top five, about 80% cheaper than our pick.
| Place | Model | Score | Price |
|---|---|---|---|
| 1 | GPT-6 AstraOpenAI | 100.0 | $10 in · $50 out |
| 2 | GPT-6.1 SolOpenAI | 99.8 | $2 in · $10 out |
| 3 | Claude Opus 5.5Anthropic | 96.3 | $4 in · $20 out |
| 4 | Claude Sonnet 5.5Anthropic | 95.2 | $2 in · $10 out |
| 5 | GPT-6 SolOpenAI | 94.1 | $2 in · $10 out |
| 6 | GPT-5.6 SolOpenAI | 93.7 | $4 in · $20 out |
| 7 | Claude Opus 5Anthropic | 91.9 | $5 in · $25 out |
| 8 | Claude Fable 5.1Anthropic | 91.8 | $10 in · $50 out |
| 9 | GPT-5.5OpenAI | 90.3 | $5 in · $30 out |
| 10 | GPT-5.6 TerraOpenAI | 89.0 | $2 in · $12 out |
Models with overlapping uncertainty share a place (shown as “=”). Prices are US dollars per million tokens from each company’s own price page.
Ranked by our Silicon score, from 8 tests by 3 publishers.