Free to start, from $8 for more
Billed monthly. Cancel any time. Limits you can always see.
Free
$0
Start without a card.
- The agent in your terminal, VS Code, JetBrains and the Mac app
- The free models
- Your own API keys and local models, with no limits from us
Plus
$8/month
Our introductory plan.
Everything in Free, plus:
- Included models: GLM, Kimi, MiniMax, DeepSeek, Qwen, Grok and Auto
- A session limit and a weekly limit, always in view
- The weekly limit resets every Thursday
Pro
$30/month
For coding every day.
Everything in Plus, plus:
- 4× Plus usage
- Claude, GPT, Gemini from credits
Compare the plans
| Feature | Free | Plus | Pro | Super | Ultra |
|---|---|---|---|---|---|
| Terminal (CLI) | |||||
| VS Code, Cursor and Windsurf | |||||
| JetBrains IDEs | |||||
| Mac app | |||||
| Free models | |||||
| Your own keys and local models | |||||
| Included models (GLM, Kimi, MiniMax…) | |||||
| Claude, GPT, Gemini (from credits) | |||||
| Usage | — | 1× | 4× | 15× | 25× |
How the limits work
- Session limit
- A session starts with your first request and lasts 5 hours. It can use up to a quarter of your weekly limit, so one long sitting never takes the whole week.
- Weekly limit
- Resets every Thursday at 00:00 UTC. Pro has four times Plus’s limits, Super 15 times and Ultra 25 times.
- Which models use them fastest
- Every request counts at its model’s price. Claude and GPT use your limits fastest; open models such as GLM, Kimi and MiniMax cost a fraction as much and last many times longer.
- What never counts
- The free models, your own API keys and local models. They keep working when a limit is reached.
Start in a few minutes
Install the CLI, sign in once, and the agent starts on Free.