Modal GPU pricing.
prices as of · 11 GPU models · 11 machines
This provider bills a GPU-hour as the unit a function is charged in, not as a machine you hold. The rate is a floor: CPU and memory are billed on top, and asking for a particular region or for your code not to be preempted costs a multiple of it. Its prices are listed and browsable here and ranked nowhere, because they are not the same purchase as a rental.
Modal rents 11 GPU models over 11 configurations, from $0.590/GPU-hr for Nvidia T4, read 17 Sept 2026.
Kind of provider
Serverless
bills the hour a function runs, not a machine you hold
Where it runs
not stated
no regions published beside the prices
Smallest billing unit
Second
billed per second, on every configuration
Spot rates
none
on-demand only
What the hourly price includes
None of Modal's 11 configurations includes a disk in the rate, and 11 state that storage is billed separately. Its largest machine is Nvidia B300, 1x B300 adding up to 288 GB of GPU memory at $7.10/hr. Egress and tax are extra unless a row says otherwise, and reserved or committed rates are not indexed anywhere on this site.
Live web · modal.com/pricing, the GPU Tasks per-second list
Boards to start with
all 11 machines ↓- H100 SXMNvidia H100 SXM5 · 80 GB per GPU · 7 providers rent it$3.9547% over the cheapestrent at Modal ↗
- A100 80GB SXMNvidia A100, 80 GB · 80 GB per GPU · 6 providers rent it$2.5080% over the cheapestrent at Modal ↗
- B200Nvidia B200 · 180 GB per GPU · 6 providers rent it$6.255% over the cheapestrent at Modal ↗
- L40SNvidia L40S · 48 GB per GPU · 5 providers rent it$1.95147% over the cheapestrent at Modal ↗
- H200 SXMNvidia H200 SXM · 141 GB per GPU · 5 providers rent it$4.5426% over the cheapestrent at Modal ↗
Everything Modal rents
Cheapest per GPU-hour first. The rate is for one GPU for one hour; the whole machine's hourly price and the cost of a gigabyte of its memory are beside it.
- Nvidia T4
1x T4 · 16 GB each · host size not stated
disk none, billed separately · $0.0369/GB-hr per GB of memory
$0.000164 a GPU-second, the preemptible region-agnostic rate · a chosen region costs 1.15x to 1.75x and non-preemptible 3x · CPU, memory and volumes billed on top · memory is the board's published spec
$0.590per GPU-hour$0.59/hr all inno spot rate
- Nvidia L4
1x L4 · 24 GB each · host size not stated
disk none, billed separately · $0.0333/GB-hr per GB of memory
$0.000222 a GPU-second, the preemptible region-agnostic rate · a chosen region costs 1.15x to 1.75x and non-preemptible 3x · CPU, memory and volumes billed on top · memory is the board's published spec
$0.799per GPU-hour$0.80/hr all inno spot rate
- Nvidia A10
1x A10 · 24 GB each · host size not stated
disk none, billed separately · $0.0459/GB-hr per GB of memory
$0.000306 a GPU-second, the preemptible region-agnostic rate · a chosen region costs 1.15x to 1.75x and non-preemptible 3x · CPU, memory and volumes billed on top · memory is the board's published spec
$1.10per GPU-hour$1.10/hr all inno spot rate
- Nvidia L40S
1x L40S · 48 GB each · host size not stated
disk none, billed separately · $0.0407/GB-hr per GB of memory
$0.000542 a GPU-second, the preemptible region-agnostic rate · a chosen region costs 1.15x to 1.75x and non-preemptible 3x · CPU, memory and volumes billed on top · memory is the board's published spec
$1.95per GPU-hour$1.95/hr all inno spot rate
- Nvidia A100, 40 GB
1x A100 40GB SXM · 40 GB each · host size not stated
disk none, billed separately · $0.0525/GB-hr per GB of memory
$0.000583 a GPU-second, the preemptible region-agnostic rate · a chosen region costs 1.15x to 1.75x and non-preemptible 3x · CPU, memory and volumes billed on top · memory is the board's published spec
$2.10per GPU-hour$2.10/hr all inno spot rate
- Nvidia A100, 80 GB
1x A100 80GB SXM · 80 GB each · host size not stated
disk none, billed separately · $0.0312/GB-hr per GB of memory
$0.000694 a GPU-second, the preemptible region-agnostic rate · a chosen region costs 1.15x to 1.75x and non-preemptible 3x · CPU, memory and volumes billed on top · memory is the board's published spec
$2.50per GPU-hour$2.50/hr all inno spot rate
- Nvidia RTX PRO 6000
1x RTX PRO 6000 · 96 GB each · host size not stated
disk none, billed separately · $0.0316/GB-hr per GB of memory
$0.000842 a GPU-second, the preemptible region-agnostic rate · a chosen region costs 1.15x to 1.75x and non-preemptible 3x · CPU, memory and volumes billed on top · memory is the board's published spec
$3.03per GPU-hour$3.03/hr all inno spot rate
- Nvidia H100 SXM5
1x H100 SXM · 80 GB each · host size not stated
disk none, billed separately · $0.0494/GB-hr per GB of memory
$0.001097 a GPU-second, the preemptible region-agnostic rate · a chosen region costs 1.15x to 1.75x and non-preemptible 3x · CPU, memory and volumes billed on top · memory is the board's published spec
$3.95per GPU-hour$3.95/hr all inno spot rate
- Nvidia H200 SXM
1x H200 SXM · 141 GB each · host size not stated
disk none, billed separately · $0.0322/GB-hr per GB of memory
$0.001261 a GPU-second, the preemptible region-agnostic rate · a chosen region costs 1.15x to 1.75x and non-preemptible 3x · CPU, memory and volumes billed on top · memory is the board's published spec
$4.54per GPU-hour$4.54/hr all inno spot rate
- Nvidia B200
1x B200 · 180 GB each · host size not stated
disk none, billed separately · $0.0347/GB-hr per GB of memory
$0.001736 a GPU-second, the preemptible region-agnostic rate · a chosen region costs 1.15x to 1.75x and non-preemptible 3x · CPU, memory and volumes billed on top · memory is the board's published spec
$6.25per GPU-hour$6.25/hr all inno spot rate
- Nvidia B300
1x B300 · 288 GB each · host size not stated
disk none, billed separately · $0.0246/GB-hr per GB of memory
$0.001972 a GPU-second, the preemptible region-agnostic rate · a chosen region costs 1.15x to 1.75x and non-preemptible 3x · CPU, memory and volumes billed on top · memory is the board's published spec
$7.10per GPU-hour$7.10/hr all inno spot rate
Frequently asked
- What kind of GPU provider is Modal?
- Serverless: it bills the hour a function runs, not a machine you hold. This provider bills a GPU-hour as the unit a function is charged in, not as a machine you hold. The rate is a floor: CPU and memory are billed on top, and asking for a particular region or for your code not to be preempted costs a multiple of it. Its prices are listed and browsable here and ranked nowhere, because they are not the same purchase as a rental.
- What is the cheapest GPU at Modal?
- Nvidia T4 at $0.590/GPU-hr, which is 1x T4 with 16 GB of memory each. The whole machine bills $0.59/hr. Read on 17 Sept 2026.
- What does Modal's hourly price include?
- None of its 11 configurations includes a disk in the hourly rate, and 11 say storage is billed separately. It publishes no region list beside its prices, so nothing here claims one. Billing is per second. Egress and tax are extra unless a row says otherwise.
- Does Modal publish a spot price?
- No. Every rate on this page is the on-demand one. Reserved and committed rates are not indexed here in any case, because a price that needs a month's commitment is not a standard recurring rate.
- How current are these Modal prices?
- Live web · modal.com/pricing, the GPU Tasks per-second list Last refresh: 17 Sept 2026.