GPU cloud · 5 GPU models tracked
packet.ai
$0.39–$6.99per GPU-hour, the cheapest row to the dearest
Two plans are priced by the hour. Dedicated gives you the whole card, exclusively yours. Dynamic shares the card with other workloads that packet.ai's scheduler places alongside yours, at roughly half the price for the same peak performance and memory. Each card is sold from one location, printed on the page beside it.
As of , packet.ai lists 5 GPU models we track, from $0.39 per GPU-hour (NVIDIA RTX 4090, on-demand) to $6.99 (NVIDIA B200, on-demand).
| Website | packet.ai/ |
|---|---|
| Pricing page | packet.ai/pricing |
| Regions and tiers | Dedicated, Dynamic — as labelled in the rates below |
| Models tracked | 5 |
Price by GPU
As of , per GPU-hour, cheapest tier first named:
- packet.ai lists the NVIDIA B200 from $6.99 per GPU-hour on-demand (Dedicated), $3.75 as community capacity (Dynamic). The on-demand rate is unchanged since first read, Sep 26, 2026. Against the market: 3% above the median rate card ($6.79 across 27 providers); 15 of the 27 providers we track list a lower on-demand rate, the lowest Deep Infra at $3.69 (Deep Infra’s rates, packet.ai vs Deep Infra).
- packet.ai lists the NVIDIA A100 80GB from $1.43 per GPU-hour on-demand (Dedicated). The on-demand rate is unchanged since first read, Sep 26, 2026. Against the market: 21% below the median rate card ($1.81 across 40 providers); 7 of the 40 providers we track list a lower on-demand rate, the lowest Deep Infra at $0.89 (Deep Infra’s rates, packet.ai vs Deep Infra).
- packet.ai lists the NVIDIA L40S from $0.92 per GPU-hour on-demand (Dedicated). The on-demand rate is unchanged since first read, Sep 26, 2026. Against the market: 34% below the median rate card ($1.40 across 28 providers); 4 of the 28 providers we track list a lower on-demand rate, the lowest Beam at $0.76 (Beam’s rates, packet.ai vs Beam).
- packet.ai lists the NVIDIA RTX PRO 6000 Blackwell from $0.66 as community capacity (Dynamic).
- packet.ai lists the NVIDIA RTX 4090 from $0.39 per GPU-hour on-demand (Dedicated). The on-demand rate is unchanged since first read, Sep 26, 2026. Against the market: the lowest on-demand rate of the 9 providers we track for this card, 13% below the median rate card ($0.45 across 9 providers).
All listed prices
One table per price type, never mixed. Every rate normalized to one GPU for one hour.
On-demand
| GPU | USD / GPU-hour | 7-day change | Type | Region | Min. commitment | Source, captured | Deploy |
|---|---|---|---|---|---|---|---|
| NVIDIA RTX 4090 | $0.39 | — | on-demand | Dedicated | — | Price page unchanged since first read, Sep 26, 2026 | Deploy ↗ |
| NVIDIA RTX PRO 6000 Blackwell | $0.66 | — | on-demand | Dynamic | — | Price page unchanged since first read, Sep 27, 2026 | Deploy ↗ |
| NVIDIA L40S | $0.92 | — | on-demand | Dedicated | — | Price page unchanged since first read, Sep 26, 2026 | Deploy ↗ |
| NVIDIA A100 80GB | $1.43 | — | on-demand | Dedicated | — | Price page unchanged since first read, Sep 26, 2026 | Deploy ↗ |
| NVIDIA B200 | $3.75 | — | on-demand | Dynamic | — | Price page unchanged since first read, Sep 27, 2026 | Deploy ↗ |
| NVIDIA B200 | $6.99 | — | on-demand | Dedicated | — | Price page unchanged since first read, Sep 26, 2026 | Deploy ↗ |
Every rate on this page as data: /providers/packet-ai/prices.json — the same fields as prices.json, narrowed to this provider. CC BY 4.0, no key, no rate limit. On your own page: embed this table (/embed/providers/packet-ai, an iframe, no script).
Compare packet.ai with
Every provider that lists at least two of the same GPUs on-demand, most shared GPUs first; each opens the side-by-side page.
- Beam 5 shared GPUs
- EmpirioLabs 5 shared GPUs
- HexGrid 5 shared GPUs
- RunPod 5 shared GPUs
- Vast.ai 5 shared GPUs
- Amazon Web Services 4 shared GPUs
- Cerebrium 4 shared GPUs
- CoreWeave 4 shared GPUs
- E2E Networks 4 shared GPUs
- Hugging Face Inference Endpoints 4 shared GPUs
- Koyeb 4 shared GPUs
- Modal 4 shared GPUs
- Oracle Cloud (OCI) 4 shared GPUs
- Verda 4 shared GPUs
- Daytona 3 shared GPUs
- Hyperstack 3 shared GPUs
- Lyceum 3 shared GPUs
- Massed Compute 3 shared GPUs
- Nebius 3 shared GPUs
- Seeweb 3 shared GPUs
- Baseten 2 shared GPUs
- Civo 2 shared GPUs
- CloudRift 2 shared GPUs
- Crusoe 2 shared GPUs
- Cudo Compute 2 shared GPUs
- Deep Infra 2 shared GPUs
- fal 2 shared GPUs
- Gcore 2 shared GPUs
- Google Cloud 2 shared GPUs
- immers.cloud 2 shared GPUs
- Jarvislabs 2 shared GPUs
- Lambda 2 shared GPUs
- Leafcloud 2 shared GPUs
- Microsoft Azure 2 shared GPUs
- Northflank 2 shared GPUs
- Nova Cloud 2 shared GPUs
- OVHcloud 2 shared GPUs
- Replicate 2 shared GPUs
- Sesterce 2 shared GPUs
Data updated: