The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Modal uses serverless, usage-based billing rather than a fixed GPU-job price. You pay for GPU time by the second, then add CPU, memory, volume storage and network egress when your workload uses them. A practical estimate is GPU rate × number of GPUs × elapsed runtime, minus any applicable plan credits, plus those supporting-resource charges.
For a single reference point, an H100 SXM5 on Modal is listed at $0.001097 per second, or about $3.9492 per hour. An eight-GPU, four-hour run would therefore incur about $126.37 in GPU charges before CPU, memory, storage and egress. Actual totals depend on the GPU model, parallelism, startup and model-loading time, container lifetime and data movement.
As an Amazon Associate I earn from qualifying purchases.
How Modal’s GPU billing works
Modal describes its platform as serverless: you request compute, run your function or job, and are metered for the resources allocated to it. Its billing documentation says you pay for the compute you use or request, with no reservations and no minimum usage-time increments. A payment method is required before using GPUs.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
GPU inference and training are billed in GPU-seconds. The basic calculation is:
#1 Best Overall
- [Ultra-Lightweight Ear Buds Designed for Small Ears] Each earbud weighs only 3.7g and features a compact, ergonomic in-ear design made especially for small ears. Secure, low-profile, and comfortable for workouts, all-day wear or overnight sleeping.The ultra-slim profile sits flush in your ear with no hard edges poking or rubbing, so you can even sleep on your side without discomfort.
- [Immersive Stereo Sound with TOZO OrigX Technology] TOZO OrigX tuning delivers clear vocals, balanced mids, and natural stereo sound for music, podcasts, and videos.
- [Long Battery Life for Daily Use] Get up to 7 hours of playtime on a single charge, with up to 32 hours total using the charging case—ideal for workdays, commuting, and extended listening sessions.
- [Bluetooth 5.3 & Stable Connection] Bluetooth 5.3 provides fast pairing, stable wireless performance, and reduced dropouts as you move around home or office.
- [Deep Bass with Clear Vocals] High-performance drivers produce punchy bass while keeping vocals clean and detailed for everyday listening.
GPU charge = price per GPU-second × GPU count × elapsed GPU runtime
That number is not necessarily the invoice total. A job can also consume physical CPU cores, RAM, persistent volumes and outbound network traffic. Those resources have their own rates. If a GPU remains allocated while a container loads a model, waits for work or performs cleanup, that elapsed allocation time can affect the bill. Modal’s own image-generation example specifically warns that model-load time, container lifetime, concurrency and resource choices can change the result.
The current rates below come from Modal’s official pricing page, accessed October 15, 2026: https://modal.com/pricing.
Recommended Free Tools
Modal GPU prices per second and per hour
Modal publishes per-second prices. The hourly figures are arithmetic conversions at 3,600 seconds per hour, not separate hourly plans or reservations.
| GPU | Per GPU-second | Approximate per hour | Typical cost question answered |
|---|---|---|---|
| Nvidia B300 | $0.001972 | $7.0992 | Highest listed rate in this table |
| Nvidia B200 | $0.001736 | $6.2496 | High-end training and inference |
| Nvidia H200 SXM | $0.001261 | $4.5396 | Large-memory accelerator |
| Nvidia H100 SXM5 | $0.001097 | $3.9492 | About $3.95 per GPU-hour |
| Nvidia RTX PRO 6000 | $0.000842 | $3.0312 | Professional workstation-class GPU |
| Nvidia A100, 80 GB | $0.000694 | $2.4984 | About $2.50 per GPU-hour |
| Nvidia A100, 40 GB | $0.000583 | $2.0988 | Lower-priced A100 option |
| Nvidia L40S | $0.000542 | $1.9512 | About $1.95 per GPU-hour |
| Nvidia A10 | $0.000306 | $1.1016 | About $1.10 per GPU-hour |
| Nvidia L4 | $0.000222 | $0.7992 | About $0.80 per GPU-hour |
| Nvidia T4 | $0.000164 | $0.5904 | Lowest listed rate in this table |
These are rates for one GPU. A four-GPU job multiplies the GPU portion by four, even if the application presents the devices as one distributed worker. For example, four H100 SXM5 GPUs running for two hours cost approximately 4 × 2 × $3.9492 = $31.5936 in GPU charges alone.
Plans, platform fees and included credits
The plan determines the platform fee, included compute credits, seat policy and listed workspace limits. It does not turn GPU usage into unlimited compute.
Rank #2
- 【IP68 Waterproof Multifunctinal Smart Watch】Torjalph fitness activity tracker smart watch is IP68 waterproof technology ,you can wear it when swimming,handwashing,it's rain-proof which means that you perform multiple exercise modes outside without worrying about sweat or rain corroding your fitness smart watch,and this watch equipped with a 230 mAh large capacity battery magnetic charging ,and this sports activity tracker watch can be used 5-7 days If fully charged.
- 【Larger 1.83'' Full Screen Watch & Multi-dial Selection】1.83 inch full HD color touch screen smart watch for men / women provides excellent 240*284 resolution image quality ,This Bluetooth smart watch offers a better touch experience and remarkable immersion and ,more than 200 watch faces for you to choose from App"Da Fit ", you can change the dial style according to your mood to wear keeping the world on your wrist
- 【Accurate Activity Tracker Smart Watch 】sports watch activity tracker uses sweat and dust resistant breathable silicone strap, our Fitness Tracker Smart Watch for Android iOS Phones supports 12 sports modes,walking,running,bicycle,swimming,badminton,basketball,football,rope skipping. This bluetooth calls smart watch for men women is lightweight and stylish, bringing more convenient and relaxing sports experience while also providing more protection.
- 【Your 24 Hour Personal Daily Health Monitor】Fitness watch for women men full touch screen IP67 waterproof bluetooth smartwatch for Android iOS Phones with heart rate blood pressure monitor multifunction Andriod smart watch with steps counter sleep monitor sports fitness watch, monitoring blood oxygen saturation allows you to understand your physical condition in real time to track fitness progress ,to know your health status in time.
- 【More Functions Bluetooth Call Smart Watch】This fitness activity smart watch full screen for iPhone and Android phones includes a bunch of new features including:The bluetooth smart watch can hang up the phone, receive and display WhatsAPP /Text /Facebook /Twitter /Messenger reminder.Support sedentary reminder, alarm clock, stopwatch, timer,weather information,music control, remote camera control compatible with Android 5.0 and above or iOS 9.0 and above smartphones.
| Plan | Platform fee | Included compute credits | Seats and limits |
|---|---|---|---|
| Starter | $0 per month | $30 per month | Up to three workspace seats; plan-specific concurrency limits apply |
| Team | $250 per month | $100 per month | Unlimited seats; higher listed limits |
| Enterprise | Custom pricing | Custom arrangement | Volume discounts and enterprise support and security features |
The Starter credit is a monthly allowance, not a cash refund. Once included credits are consumed, additional usage is billed according to the applicable resource rates. The Team plan’s $250 platform fee is separate from its $100 monthly compute credit. Enterprise pricing must be obtained from Modal; public pricing does not establish a universal enterprise rate.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Clear out junk files and repair common Windows errors3Scan for outdated or missing drivers - takes under a minuteCPU, memory, storage and network charges
GPU time is only one part of a serverless workload. Modal lists the following supporting-resource prices:
| Resource | Rate | How it is measured |
|---|---|---|
| CPU | $0.0000131 | Per physical core-second |
| Memory | $0.00000222 | Per GiB-second |
| Volumes | $0.09 | Per GiB-month |
| Network egress | $0.04 | Per GiB transferred out |
Modal also lists free allowances, including up to 1 TiB per month of volume storage and plan-specific network-egress allowances. The exact allowance depends on the plan and should be checked on the pricing page before budgeting. Free allowances do not remove the need to account for unusually large datasets, model artifacts or outbound results.
How much does Modal inference cost?
Custom GPU inference
With a custom inference stack, you select the GPU and are charged for the runtime of that stack. A first-pass estimate is:
requests × average allocated seconds per request × GPU rate
Then add CPU and memory time, volume usage and egress. Measure allocated time rather than only model execution time: image decoding, batching, model loading and shutdown may all occur while the container is using resources.
Rank #3
- [Low Noise High Fidelity] New bee guitar cord with 22AWG oxygen-free-copper center conductors provides enhanced signal transfer, the PVC insulation material and the high-density stranded copper double shield provide noise rejection for quiet operation
- [Ideal for Choice] New bee 1/4 straight and right angle metal connectors, ensures a secure connection in any instrument, pedal, or amp jack. L shape great fro on stage guitar stands
- [Wide Compatible] Suitable for mixer, power amplifier, microphone, effector, decoder, equalizer, electoronic organ, electric guitar, electric bass and other electronic equipments. Widely used in the installation of smart home, karaoke speaker, professional stage audio, etc
- [Easy Disassemble] Making incidental repairs quick and easy! Very nice, flexible cable
- [Dont Hesite to Order] 24 hours customer services and professional technology team are standing by
Modal’s Stable Diffusion example
Modal gives a concrete example for a Stable Diffusion task on an A10G. It reports about 1.5 seconds per image and approximately $0.000458 in GPU charges, $0.000020 in CPU charges and $0.000013 in memory charges. The stated total is approximately $0.000491 per image, or $0.491 per 1,000 images.
This is an illustrative configuration, not a guaranteed price for every A10G deployment. Different image sizes, model weights, startup behavior, concurrency, container lifetime and allocated CPU or memory can change both latency and cost. The A10G example should not be substituted for the separately listed Nvidia A10 rate without measuring your own stack.
Shared Endpoints
Modal also offers Shared Endpoints with token-based pricing. Token usage is billed from the first request and is separate from the Starter plan’s included compute credit. This is a different billing model from running a custom GPU container: estimate Shared Endpoint spend from the endpoint’s published token rate and your input and output token volume, rather than multiplying a GPU-hour price. Product details are available at https://modal.com/products/inference.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →How much does Modal training cost?
Training and fine-tuning use the same serverless billing model. There is no single “training job” price because the result depends on:
- GPU model and number of GPUs
- Elapsed runtime, including distributed startup and checkpoint phases
- CPU cores and memory allocated to each worker
- Persistent volume capacity and retention time
- Data uploaded to or downloaded from the environment
- Whether the job runs continuously or is split into multiple launches
Use this estimate:
training total ≈ GPU rate × GPU count × elapsed hours × 3,600 + CPU + memory + storage + egress
Worked GPU-only training estimates
| Configuration | GPU-only calculation | GPU charge |
|---|---|---|
| One H100 SXM5 for 8 hours | 1 × 8 × $3.9492 | $31.5936 |
| Eight H100 SXM5 GPUs for 4 hours | 8 × 4 × $3.9492 | $126.3744 |
| Four A100 80 GB GPUs for 10 hours | 4 × 10 × $2.4984 | $99.936 |
| One L4 for 24 hours | 1 × 24 × $0.7992 | $19.1808 |
These figures deliberately exclude CPU, RAM, volume and egress. Distributed training can also spend time synchronizing workers or recovering from failed attempts, so benchmark a representative run and include checkpoints and validation in the elapsed time.
Rank #4
- Digital Family Calendar: BSIMB digital calendar wall touch screen puts an end to your chaotic life. Designed for busy families, the digital calendar keeps you organised and in step with your family. Assign different colours and tasks to each family member to make family scheduling easier and more intuitive.
- Easy Setup and Auto-Sync: Plug in, connect to WLAN and auto-syn your electric calendar in minutes. Digital wall calendar compatible with Google calendar, iCloud calendar, Outlook calendar, Cozi calendar and Yahoo calendar. Add events directly on the device or through the ecalendar app. Electronic calendar stays current whether you're at work or traveling—no manual updates needed.
- All-In-One Home Assistant: Smart calendar builds habits and encourage independence with personalised chores. Display daily meal plan to reduce the daily ‘What's for dinner?’ stress. To-do lists keep track of everything you want. Turning chores into valuable achievements with stars and rewards, making life easier and more enjoyable.
- Manage Anytime Anywhere: BSIMB electronic calendar for family stay connected with the powerful ecalendar app. Access and manage your digital wall calendar anytime, anywhere with free features like stars & rewards, photo screensaver, chore charts, meals plans, to-do lists etc. Electronic planner paid features like Magic Import tailor advanced tools for users with special needs.
- Stay in Touch with Loved Ones: Share photos or videos from your phone to electronic wall calendar with app, instantly update your life moment, and keep in touch with your family, especially those who are far away; you can also surprise your smart calendar digital wall by sending him a wish card; it's also great for keeping tech-illiterate elders in touch with the youngsters.
A step-by-step way to forecast your bill
- Choose a GPU based on memory and throughput. Record the exact model, such as H100 SXM5 rather than a generic “H100,” because Modal prices models separately.
- Measure allocated runtime. Run a representative inference batch or training interval and include startup, model loading, evaluation, checkpointing and teardown when those happen inside the allocation.
- Multiply GPU usage. Convert hours to seconds or use the hourly arithmetic above, then multiply by the number of simultaneously allocated GPUs.
- Add supporting resources. Estimate physical core-seconds, GiB-seconds, stored GiB-months and outbound GiB using Modal’s published rates.
- Apply the plan allowance. Subtract only the applicable Starter or Team compute credit from eligible compute charges; keep the Team platform fee in the monthly total.
- Set a budget before scaling. Use workspace-level and environment-level budgets to limit unexpected spend, then review actual usage after the first production run.
For a production forecast, calculate a low, expected and peak case. The peak case should reflect maximum concurrency, longest model-load time and the largest expected output transfer.
Billing mechanics and spend controls
Modal’s billing documentation says workspaces are billed monthly, with incremental charges when usage crosses specified thresholds. Billing is workspace-based, so teams should decide which projects share a workspace before assigning budgets. Workspace-level and environment-level budgets provide guardrails for experiments and deployments.
There are no reservations and no minimum usage-time increments, so you do not have to commit to a month-long GPU allocation or a minimum block of hours. That flexibility is useful for bursty inference and short training experiments, but a workload that repeatedly keeps containers alive can still accumulate runtime charges.
See the billing documentation for current account and threshold details: https://modal.com/docs/guide/billing. Modal summarizes the model as “you always pay for what you use and nothing more,” while the practical meaning is that every metered resource and allocated second should be included in your estimate.
Common cost surprises and how to prevent them
Confusing GPU execution time with end-to-end allocation time
A fast kernel does not guarantee a cheap request if the worker spends substantial time loading weights or waiting between batches. Keep containers warm only when the latency benefit justifies the additional allocated time, and measure both cold and warm paths.
Multiplying by the wrong GPU count
Distributed training and tensor-parallel inference charge for every allocated device. Confirm whether a job requests one multi-GPU worker or several workers, then multiply the rate by the actual device count.
Ignoring data movement
Large checkpoints, generated media and exported datasets can create egress charges even when GPU time is modest. Keep frequently reused data in the appropriate volume and estimate outbound GiB for every destination.
Assuming a free credit covers the whole invoice
Starter and Team credits are monthly compute credits, while platform fees, resource categories and usage beyond the allowance remain separate considerations. Track the credit balance and set environment budgets for experiments.
Using a benchmark as a guarantee
The Stable Diffusion figure is one documented configuration. Your model architecture, precision, batch size, concurrency and container settings can produce a different cost per request. Re-run the estimate after changing any of those variables.
Is Modal pay-as-you-go?
Yes. Modal’s serverless model is pay-as-you-go for metered compute, with monthly workspace billing and no reservation requirement or minimum usage-time increment. The predictable part is the published resource rate; the variable part is how long and how often your application allocates each resource. For procurement, Modal is also listed on AWS Marketplace as an enterprise purchasing option, with buyers directed to contact Modal for pricing: https://aws.amazon.com/marketplace/pp/prodview-j727623xqhh2k.
What to record before approving a GPU budget
- Exact GPU model and number of concurrent GPUs
- Cold-start and warm-request duration
- Average and peak request volume, or expected training hours
- CPU cores and memory per worker
- Volume size and retention period
- Expected egress in GiB
- Plan fee and remaining monthly compute credit
- Workspace and environment budget limits
With those inputs, Modal’s published rates are sufficient to produce a defensible estimate for both inference and training. The final invoice will follow actual allocation time and metered resource consumption, not a fixed per-job quote.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
