Curated models for your best work
Join the waitlistAll models
Compare models
List price per 1M tokens. The best value in a row is bold; the rest are dimmed.
Scroll sideways to see every model.
| Attribute | AnthropicRemove | AnthropicRemove | OpenAIRemove |
|---|---|---|---|
| Price per 1M tokens | |||
| Input | $4.00 | $2.00 | $2.00 |
| Output | $20.00 | $10.00 | $10.00 |
| Cached input | $0.20 | $0.20 | $0.10 |
| Long prompts | - | - | $4.00 in, $15.00 out past 272K |
| Limits | |||
| Context window | 1M | 1M | 1.05M |
| Max output | 128K | 128K | 128K |
| Scores | |||
| Intelligence | 169 | 166 | 168 |
| Coding | 64.8% · Claude Code | 61.8% · Claude Code | 58.2% · Codex |
| Capabilities | |||
| Reasoning | Yes | Yes | Yes, 4 effort levels |
| Image input | Yes | Yes | Yes |
| Tool calls | Yes | Yes | Yes |
| Structured outputs | No | No | Yes |
| Open weights | No | No | No |
| Data | |||
| Where prompts go | The provider's own retention policy applies. | The provider's own retention policy applies. | The provider's own retention policy applies. |
| Release | |||
| Released | Sep 22, 2026 | Sep 28, 2026 | Sep 29, 2026 |
Intelligence is the Epoch Capabilities Index (Epoch AI, CC BY 4.0). Coding is Terminal-Bench 4.0, the share of tasks solved by the agent shown.
TokenSpend stores no prompt or response content for any model. See Trust.