How Billing Works
There are two cost streams, both drawing from a single credit pool:- Compute — billed per minute of sandbox runtime. Every active session consumes credits while it runs.
- LLM — billed per model inference. Token usage from all LLM calls is tracked and converted to credits automatically.
Both compute and LLM costs deduct from the same credit balance. There are no separate budgets to manage.
Plans
| Plan | Monthly | Credits Included | Max Sessions | Max Snapshots |
|---|---|---|---|---|
| Dev | $20 | 1,000 | 10 | 5 |
| Pro | $500 | 7,500 | 100 | 200 |
Trial Credits
New organizations receive 1,000 trial credits at signup — no payment method required. This is enough to explore the platform, run sessions, and see results before committing to a plan.Usage Monitoring
The billing dashboard shows real-time spend per organization, broken down by compute and LLM costs. You can track how credits are being consumed across sessions and models.Auto-Pause
When your credit balance reaches zero, active sessions are automatically paused and snapshotted. Your work is preserved — nothing is lost. Once credits are added (via a plan or top-up), paused sessions can be resumed from exactly where they left off.Sessions are paused, not terminated. All state is saved in a snapshot so you can pick up where you left off.