Pay per minute. Stop paying when you stop.
Neural Studio Pro is $29 a month. Every GPU below is listed at its starting price from the live rate card. Notebooks, files, experiment tracking, and collaborators are included.
How billing works
- Pro is $29 a month, and every account needs it. 1,000 credits land each cycle and expire at its end. Cancel any time; Pro stays active until the end of the period you paid for.
- $1 = 100 credits. One credit is one cent of compute; top up in packs from $5, valid for 12 months.
- Per-minute, from running to stopped. The meter starts the moment your runtime is running and stops when you stop it. Provisioning time is never charged.
- A stopped runtime costs nothing. Your workspace files stay put in your 250 GB storage pool, shared by all your workspaces; only running compute burns credits.
- Spot is cheaper, and interruptible. Where a provider sells spot, the launch picker shows its spot rate; the provider can reclaim a spot runtime, so keep checkpoints in your workspace.
GPU pricing
| GPU | VRAM | Starting from |
|---|---|---|
| T4 | 16 GB | from 52.6 cr/hr$0.53/hr per GPU |
| RTX4000Ada | 20 GB | from 76 cr/hr$0.76/hr per GPU |
| L4 | 24 GB | from 80.5 cr/hr$0.80/hr per GPU |
| A10G | 24 GB | from 100.6 cr/hr$1.01/hr per GPU |
| A6000 | 48 GB | from 109 cr/hr$1.09/hr per GPU |
| A10 | 24 GB | from 129 cr/hr$1.29/hr per GPU |
| RTX6000Ada | 48 GB | from 142 cr/hr$1.42/hr per GPU |
| A100 | 40 GB | from 157 cr/hr$1.57/hr per GPU |
| L40S | 48 GB | from 157 cr/hr$1.57/hr per GPU |
| RTX6000 | 96 GB | from 180 cr/hr$1.80/hr per GPU |
| A100-80GB | 80 GB | from 213.6 cr/hr$2.14/hr per GPU |
| H100 | 80 GB | from 329 cr/hr$3.29/hr per GPU |
| H200 | 141 GB | from 447 cr/hr$4.47/hr per GPU |
| B200 | 180 GB | from 669 cr/hr$6.69/hr per GPU |
| B300 | 268 GB | from 1,069.9 cr/hr$10.70/hr per GPU |
| CPU only | — | from 11.9 cr/hr$0.12/hr |
GPU prices are per GPU-hour: the lowest on-demand rate for that GPU across our cloud providers, tracking their live list prices and refreshed at most every minute. Multi-GPU machines bill as a whole. The launch picker shows the exact rate for every machine size, and the spot rate where one is offered; that is the rate you are charged.