# Modal Pricing

**Source:** https://modal.com/pricing — **Snapshot:** 2026-08-12 (page lists per-second rates; converted to $/hr)
**Billing:** serverless — you pay **only while a container is running**. No idle cost, no destroy step, scale-to-zero. Free tier: $30/month credit; volume pricing at $250+/month.

| GPU | $/hr (on-demand, converted from per-second) |
| --- | ---: |
| T4 | 0.59 |
| L4 | 0.80 |
| A10 | 1.10 |
| L40S | 1.95 |
| A100 40GB | 2.10 |
| A100 80GB | 2.50 |
| RTX PRO 6000 | 3.03 |
| H100 SXM5 | 3.95 |
| H200 SXM | 4.54 |
| B200 | 6.25 |
| B300 | 7.10 |

Notes:
- The killer property is **zero idle cost**: for sparse batches (a job every few hours), Modal can beat a rented idle instance even at a higher hourly rate — the gpu-ops keep-alive math decides.
- The counterweight is **cold starts**: each scale-up pays image pull + model load before billing even starts. For a dense queue, a kept-alive Vast 4090 at $0.14/hr dominates.
- Spot (cheaper) capacity exists for the same GPUs on the pricing page — the snapshot above is on-demand.
- 80GB-class cards (H100/H200/B200/B300) are the ones that matter for big video models; the 24GB tier (L4) handles Wan 2.2 5B.

**Verify live before committing spend** — this is a snapshot.