# Prime Intellect Pricing

**Source:** https://www.primeintellect.ai (marketplace listings) — **Snapshot:** 2026-08-12
**Billing:** decentralized marketplace — dynamic per-cluster pricing, pay per GPU-hour, spot available. No published static rate card; prices are set by cluster owners and you can bid or take the ask.

Example marketplace listings at snapshot time:

| GPU | $/hr (example listings) |
| --- | ---: |
| H200 | 0.47 – 3.14 (cluster-dependent) |
| H100 | 2.43 (spot from 0.94) |
| B200 | 3.49 |
| B300 | 4.99 |

Notes:
- The H200 range shows how cluster-dependent pricing is — same GPU, 6x spread across hosts. Shop per batch, not per provider.
- **Spot exists** (H100 spot ≈ 60% below on-demand in the example) — for retryable video batches that's the interesting lane.
- Bidding: you can submit parallel bids for options (H100/H200/B200/B300) and take the first fill.
- CLI surface changes frequently — verify commands against docs.primeintellect.ai at deploy time before scripting anything.

**Verify live before committing spend** — this is a snapshot of example listings, not a rate card.