Model rates.
How quickly each model choice uses the usage your plan includes.
Your plan includes an amount of AI usage each month, on every plan, including annual plans. The model you pick decides how fast it is used. Auto is the recommended value setting: it draws on a range of models and picks the one best suited to each task, at one steady rate. Pick a Named model instead and it does more per turn, and uses your included usage faster — paid from the same included usage, at that model’s own published rate.
| Model | Input | Output |
|---|---|---|
AutoRecommended | $1.5 | $5.0 |
Claude Sonnet 5Named | $2.0 | $10.0 |
Claude Opus 5Named | $5.0 | $25.0 |
Claude Fable 5Named | $10.0 | $50.0 |
OpenAI GPT-5.6 SolNamed | $2.5$5.0 over 272K tokens | $15.0$22.5 over 272K tokens |
Google Gemini 3.1 ProNamed | $2.0$4.0 over 200K tokens | $12.0$18.0 over 200K tokens |
xAI Grok 4.6Named | $2.0$4.0 over 200K tokens | $6.0$12.0 over 200K tokens |
Qwen3.8 MaxNamed | $2.0 | $6.0 |
Kimi K3Named | $3.0 | $15.0 |
Rates shown per 1M tokens.
Rates, models, names, units, and availability may change going forward. The rate in effect at the time you use a model is the rate that governs that usage.
Usage you have already run is never recalculated at a new rate. Any change to your subscription price is handled through the subscription notice path — never as a silent charge after the fact.