With Modal, you always pay for what you use and nothing more. You never pay for idle resources — just actual compute time, by the CPU cycle.
Physical core (2 vCPU equivalent)
*minimum of 0.125 cores per container
*includes 1 TiB / mo free
Built for small teams and independent developers looking to level up.
Built for startups and larger organizations looking to scale quickly.
For organizations prioritizing security, support, and everlasting confidence.
Early-stage startups can get free compute credits on Modal.
Graduate students, labs, and researchers can get up to $10k free compute credits on Modal
Transact through the AWS and GCP marketplace to use committed spend on Modal.
Only pay for what you use. Burst up to what you need without over-allocating CPU or memory in advance.
CPU
Physical core (2 vCPU equivalent)
$0.00003942 / core / sec
*minimum of 0.125 cores per container
Memory
$0.00000667 / GiB / sec
Modal is serverless, which means that we instantly autoscale up and down for you based on request volume. For spiky or unpredictable workloads, we are more cost-effective than fixed on-demand/reserved compute.
You’re billed for reserved GPUs, even when usage drops.
$5,400
75 GPUs * 24 hrs * $3 / GPU-hr
Only pay for what you use, when you use it.
$4,740
Avg 50 GPUs * 24 hrs * $3.95 / GPU-hr
| Plans | Starter: $0 + compute / month | Team: $250 + compute / month | Enterprise: Custom compute |
|---|---|---|---|
Workspace | |||
| Number of seats | Up to 3 | Unlimited | Unlimited |
Credits and variable costs | |||
| Included compute | $30 / month | $100 / month | Custom |
| Included network egress | 1 TiB / month | 10 TiB / month | 100 TiB / month |
Features | |||
| Containers | 100 | 5000 | Custom |
| GPU concurrency | 10 | 50 | Custom |
| Deployed apps | 200 | 1000 | 1000 |
| Log retention | 1 day | 30 days | Custom |
| Custom domains | Not included | Included | Included |
| Real-time metrics | Included | Included | Included |
| Sharing and collaboration | Included | Included | Included |
| Secrets | Unlimited | Unlimited | Unlimited |
| Custom images | Unlimited | Unlimited | Unlimited |
| Instant deploys | Unlimited | Unlimited | Unlimited |
| Cron jobs | 5 deployed crons | Unlimited | Unlimited |
| Deployment rollbacks | Not included | 3 versions | Custom |
| Environment-level budgets | Not included | Included | Included |
| Distributed queue | Unlimited | Unlimited | Unlimited |
| Distributed dict | Unlimited | Unlimited | Unlimited |
| Region selection | 1.15 - 1.75x base prices | 1.15 - 1.75x base prices | 1.15 - 1.75x base prices |
| Non-preemptible execution | 3x base prices | 3x base prices | 3x base prices |
| Cloud Marketplaces | Not included | Not included | Use committed spend on AWS and GCP |
| Support | Access to Modal Community Slack | Access to Modal Community Slack | + Support via private Slack + Embedded ML engineering services |
Security | |||
| SOC 2 compliance | Included | Included | Included |
| HIPAA compatibility | Not included | Not included | Included |
| Audit logs | Not included | Not included | Included |
| Static IP proxy | Not included | Included | Included |
| RBAC | Not included | Included | Included |
| SSO | Not included | Not included | Included |
Modal bills you during application load time and during the time your container is processing inputs. By default, we keep containers alive for 60 seconds after the last inputs before we shut them down, which we also bill for. However, this timeout length is configurable. Once your application has scaled down to zero containers you no longer get charged.
Modal continuously records CPU and memory consumption. CPU usage is tracked in terms of fractional physical cores, allowing for higher utilization compared to providers that set minimum thresholds for core charges. Note that each physical core is equivalent to 2 vCPUs as offered by AWS or GCP.
You'll be charged based on whichever is higher: the amount you request or the amount you actually use. Each container has a minimum request of 128 MiB of memory and 0.125 CPU cores, which you can increase if you need more resources.
Modal's serverless pricing can't be directly compared to traditional on-demand or spot instance pricing.
Traditional cloud compute doesn't have instant spin-up times. You might spend 10+ minutes provisioning an instance and another 5 terminating it. You also have to commit to a minimum usage time per instance, usually between 1 minute and 1 hour. All this time is billable. In contrast, Modal spins containers up and down in 1-2 seconds. You don't commit to any minimum usage time per instance.
Even though our per-hour prices may sometimes appear higher, your overall costs are lower if your request volumes are spiky or unpredictable. Modal autoscales fast, so you pay for significantly less non-execution time versus traditional compute. You also save on cost by not having to over-provision compute to handle peak loads.
You schedule a job to run once a day for an hour. The job uses 50% of one CPU core and 256 MiB of memory.
This job will cost $0.02365/day for the CPU, and $0.00200/day for the memory, adding up to $0.02565/day, or $0.77 over 30 days.
You run stable diffusion on an A10G GPU. This will run for about 1.5 seconds to generate each image, while using 4 GiB RAM and 1 CPU.
This will cost approximately $0.000458/image in GPU charges, $0.000020/image in CPU charges, and $0.000013/image in memory charges, adding up to $0.000491/image in total charges, i.e., $0.491 per 1,000 images.
Workspaces receive a free allotment of compute per month, but usage on the Starter and Team plans is pay-as-you-go and does not require pre-purchasing compute or tokens. If you want to limit your monthly spend, you can set a Workspace budget.
The starter plan allows up to three users in a workspace. Upgrading to the team plan unlocks an unlimited number of seats.
You can deploy any application that can be containerized, using Python. Modal is especially well-suited for compute-intensive workloads that benefit from horizontal scaling. This includes ML inference at scale, model training & fine-tuning, data pipelines, numerical & scientific computing, compute-heavy job queues, and much more.
At the moment, credits from AWS, GCP, or Azure cannot be applied towards your Modal usage. However, we have a special partnership with AWS that allows eligible startups to unlock additional Modal credits if they are part of AWS Activate. To find out if your company is eligible, please apply to our startup credits program.
Included usage applies to compute: Functions, Sandboxes, and Notebooks, billed by the second for CPU, memory, and GPU time. Volumes include up to 1 TiB free.
It does not apply to Shared Endpoints, which are billed per token rather than per second. Token usage is always billed, starting from your first request. To cap out-of-pocket charges, see spend limits.
Starter has no monthly fee and includes $30 of compute (GPUs, CPUs, Memory) per month. You're billed when you exceed that, or when you use something the included compute doesn't cover.
Free usage of Shared Endpoints is not part of the Starter plan. Tokens are billed from your first request, so they will appear on your invoice even if you haven't used your full $30 of included compute. This took effect September 1, 2026.
$30 / month free compute
