Pay only for the GPU time your agents use. Every plan includes prepaid credits and per-second billing with no minimums.
5 concurrent runs, $100 per-run cap, $30/mo included credits
100 concurrent runs, $200 per-run cap, $500/mo included credits, dedicated support
Custom limits, SLA, SSO, dedicated infrastructure. Contact us.
Top up your Vault with prepaid GPU credits. Larger packs include a discount.
| Feature | Free | Pro | Team Starter | Team |
|---|---|---|---|---|
| Concurrent GPU runs | 1 | 3 | 20 | 50 |
| Per-run cap | $5 | $25 | $25 | $50 |
| Included credits | $5 once | $10/mo | $50/mo | $150/mo |
| Multi-user workspace | — | — | ✓ | ✓ |
| Slack notifications | — | — | ✓ | ✓ |
| BYO RunPod | — | — | — | ✓ |
| Per-agent GPU budgets | — | — | ✓ | ✓ |
| Audit log | — | — | ✓ | ✓ |
You're charged only for the actual GPU time your code uses. If your run takes 3.7 seconds on an RTX 4090, you pay for exactly 3.7 seconds. No rounding up to the nearest minute.
GPU runs will fail with a clear error message returned to your agent. Your agent can then prompt you to top up. We never charge your card without explicit action.
Yes, on the Team plan and above. Connect your RunPod API key in workspace settings. We handle governance and attribution; you pay RunPod directly for compute.
RTX 4090 (24GB), A100 80GB, and H100 SXM (80GB). More GPUs are added based on demand.