H100 GPU sandboxes start at $2.95/hr on Thunder Compute and cost up to $3.95/hr on Modal and Daytona's on-demand tier. Daytona also offers cheaper preemptible rates, but sandboxes can be terminated without notice. All three platforms charge separately for vCPUs and memory; GPU rates are only part of the bill.
This guide breaks down the GPU sandbox bill and the rules that affect the total. Base prices were last reviewed in October 2026, before volume discounts or region multipliers.
Takeaways
- vCPU and memory add $0.92-1.34/hr to a sandbox for 8 vCPUs and 32 GiB of memory.
- Full configurations at base rates range from $3.20/hr (Daytona preemptible) to $5.29/hr (Modal).
- Per second billing is the standard among covered providers.
- Preemptible GPU sandboxes cost less but can be terminated without notice.
- Modal with narrow region selection costs 2.4x more than Thunder Compute for the example configuration ($9.25/hr vs $3.88/hr).
Environments for agentic workloads
Non-preemptible GPU sandboxes, from $1.95/hr.
GPU Sandbox Pricing at a Glance
| Rate (per hour) | Thunder Compute | Daytona | Modal |
|---|---|---|---|
| H100 GPU | $2.95 on-demand | $2.27 preemptible, $3.95 on-demand | $3.95 preemptible |
| A100 GPU 80GB | $1.95 on-demand | - | $2.50 preemptible |
| vCPU/hr | $0.051 | $0.050 | $0.0711 |
| Memory, per GiB/hr | $0.016 | $0.016 | $0.024 |
| Storage, per GiB/hr | $0.00012 | $0.000108 after the first 5 GiB | Not billed (Sandbox disk size can't be increased) |
| Billing increment | Per second | ||
| Region Multipliers | None | None | 1.15x for broad selection, 1.75x for narrow2 |
| Egress, per GiB | Not billed | Not billed4 | $0.04 after the included allowance3 |
| 1 Modal's Sandbox rates are $0.1419/physical core/hr (1 physical core = 2 vCPU). 2 Applies only when you pin sandboxes to a region. 3 1TiB/month on Starter, 10TiB/month on Team and 100TiB/month on Enterprise, shared across the workspace. 4 Daytona bills only reserved vCPU, RAM and disk, plus GPUs, with no egress charge. |
|||
What a GPU Sandbox Actually Costs per Hour
A realistic H100 GPU sandbox costs $3.20-$5.29/hr across these options once CPU and memory are included. This cost table prices one H100 with 8 vCPUs, 32 GiB of memory and 100 GiB of storage at list rates.
| Option | H100 GPU | 8 vCPU | 32 GiB Memory | 100 GiB Storage | Total per Hour |
|---|---|---|---|---|---|
| Daytona - preemptible | $2.27 | $0.40 | $0.52 | $0.01 | $3.20 |
| Thunder Compute | $2.95 | $0.41 | $0.51 | $0.01 | $3.88 |
| Daytona - on-demand | $3.95 | $0.40 | $0.52 | $0.01 | $4.88 |
| Modal | $3.95 | $0.57 | $0.77 | $0.001 | $5.29 |
| Modal - Narrow Region | $3.95 | $0.57 | $0.77 | $0.001 | $9.252 |
| 1Modal doesn't bill Sandbox disk separately. 21.75 multiplier applies for narrow region selection |
|||||
Daytona preemptible sandboxes have the lowest hourly rate if runs survive. Evicted runs must be paid for again. At base rates, Thunder Compute is about 20% cheaper than Daytona on-demand and 27% cheaper than Modal for this configuration.
GPU Rates for Other GPUs on Modal and Daytona
Modal and Daytona charge identical rates for every GPU but under different conditions: Daytona's GPU sandboxes are on-demand, and Modal's are preemptible.
| GPU | Daytona Preemptible | Daytona On-Demand | Modal |
|---|---|---|---|
| B300 | $4.08 | $7.10 | $7.10 |
| B200 | $3.59 | $6.25 | $6.25 |
| AMD MI355X | $3.44 | $5.99 | - |
| H200 | $2.61 | $4.54 | $4.54 |
| RTX PRO 6000 | $1.74 | $3.03 | $3.03 |
| A100 40GB | - | - | $2.10 |
| L40S | - | - | $1.95 |
| RTX 5090 | $0.74 | $1.29 | - |
| A10 | - | - | $1.10 |
| RTX 4090 | $0.57 | $0.99 | - |
| L4 | - | - | $0.80 |
| T4 | - | - | $0.59 |
Additional Billing
Three additional items beyond the GPU have an impact on the bill:
- Resources billed on top of the GPU.
- Preemption.
- Egress.
Thunder Compute, Daytona and Modal handle each rule differently.
CPU, Memory and Storage on Top of the GPU
All three platforms bill GPU, CPU and memory for as long as a sandbox runs. Thunder Compute and Daytona also bill storage; Modal does not bill Sandbox disk separately. On Thunder Compute and Daytona, 8 vCPUs and 32 GiB of memory add $0.92/hr to an H100 sandbox. On Modal, the same resources cost $1.34/hr.
Resource sizing matters as much as the GPU choice. Thunder Compute and Daytona bill the vCPUs and memory you configure. Modal bills whichever is higher, your request or actual usage.
Modal Sandboxes Pricing Breakdown
Modal Sandboxes pricing has two easy-to-miss details:
- Sandbox CPU costs roughly 3x Modal's function rate: $0.00003942/physical core/second, where one physical core equals 2 vCPUs.
- Pinning sandboxes to a region multiplies prices by 1.15x for broad region selections like "us" and 1.75x for narrow ones like "us-west" (see region selection documentation).
Modal GPU Sandboxes use Modal's standard GPU rates, and are preemptible only.
Daytona Sandbox Pricing: On-Demand vs Preemptible
Daytona's preemptible GPU rate is about 42% lower than its on-demand rate for every GPU. A Daytona H100 costs $3.95/hr on-demand and $2.27/hr preemptible. Preemptible sandboxes can be terminated without notice, keeping evicted sandboxes retrievable for 24 hours.
Daytona tiers also behave differently when GPUs are out of stock:
vCPU, memory and storage cost the same on both Daytona tiers.
Egress: Modal's New Network Charge
Modal started charging for network egress on October 1, 2026. Modal's egress billing charges $0.04/GiB once a workspace uses its allowance for the billing cycle:
- Starter: 1 TiB included
- Team: 10 TiB included
- Enterprise: 100 TiB included
Egress covers traffic leaving a container's network interface, including private traffic to other Modal containers and uploads through Cloud Bucket Mounts. Reads from and writes to Modal Volumes don't count.
Idle Time: Timeouts and Auto-Stop Defaults
GPU sandboxes bill for the whole GPU while they exist, whether they are busy or not. A GPU that works for 10% of a session pays 10x of what it actually needs. The rest of the session usually goes to LLM calls and tool results.
Default timeouts cap how long an idle sandbox keeps billing:
- Thunder Compute sandboxes expire after 300 seconds by default. This timeout can be extended or disabled.
- Modal Sandboxes default to 5 minutes and can run for up to 24 hours.
- Daytona sandboxes auto-stop after 15 idle minutes by default.
Set explicit timeouts that match the workload rather than relying on platform defaults.
Worked Example: 1,000 Coding-Agent Eval Runs
Running 1,000 evals at 20 minutes each comes to about 333 GPU-hours. The eval cost table uses the same H100, 8 vCPU, 32 GiB configuration as the hourly cost table.
| Option | Cost per Hour | 1,000 20-Minute Runs |
|---|---|---|
| Daytona preemptible | $3.20 | $1,0651 |
| Thunder Compute | $3.88 | $1,292 |
| Daytona on-demand | $4.88 | $1,625 |
| Modal | $5.29 | $1,7621 |
| 1Assumes no runs are evicted; every eviction adds the cost of a rerun. | ||
Thunder Compute saves about $333/month against Daytona on-demand and $470/month against Modal for 1,000 eval runs, with no evictions to bloat costs or complicate implementation.
When Daytona Preemptible Sandboxes Cost Less
For this workload, Daytona preemptible costs less only if:
- Evictions waste less than 71 GPU-hours.
- Your eval harness retries interrupted runs to avoid engineering costs.
Daytona doesn't publish eviction rates, so teams have to measure lost GPU-hours on their own workloads. The 71 GPU-hour budget assumes an evicted run restarts from scratch and is billed until eviction. Reruns also delay results, and preemptible requests fail immediately when no spot capacity is available.
Run evals on non-preemptible H100 sandboxes on Thunder Compute.
Cheapest GPU sandbox for AI Agents: Which Option Fits Your Workload?
The cheapest GPU sandbox for AI agents depends on job length and the cost of an interruption. Sandbox pricing for AI agents only makes sense alongside those two factors.
Short, Interruptible Jobs
Daytona's preemptible tier is the cheapest option for short tasks that can be retried. When a task takes a few minutes and the harness retries on failure, an occasional eviction costs little.
Long Evals and RL Runs
Thunder Compute is the cheapest non-preemptible H100 of the three. On long evals and RL rollouts, an eviction throws away the work already done. A run guaranteed to finish at $2.95/hr can end up cheaper than a lower rate plus reruns, and the gap grows with the eviction rate.
Read nextBest GPU Sandboxes for AI Agents Features, not just price, in our ranking of the best GPU sandboxes.
Free Credits and Startup Programs
Daytona advertises $200 in free compute, but free credits can't be used for GPU sandboxes and two new accounts we created in October 2026 each received $100.
Daytona's startup program grants $10,000 in credits on acceptance, with up to $50,000 at Daytona's discretion.
Modal's Starter plan includes $30/mo of compute credit after adding a valid payment method.
Last Thoughts on GPU Sandbox Pricing
GPU sandbox prices look similar at the headline rate and diverge once CPU, memory, preemption and idle time are added. A typical H100 sandbox costs about $3.88/hr on Thunder Compute, $4.88/hr on Daytona on-demand and $5.29/hr on Modal.
Price the full configuration you plan to run, then weigh what an interrupted run would cost before choosing a preemptible tier.
FAQ
How Much Does a GPU Sandbox Cost per Hour?
An H100 GPU sandbox (8 vCPU, 32 GiB configuration) costs $3.88/hr on Thunder Compute and $5.29/hr on Modal.
How Much Does Daytona Charge for GPU Sandboxes?
Daytona charges $3.95/hr for an on-demand H100 and $2.27/hr preemptible, plus $0.0504/vCPU-hr and $0.0162/GiB-hr of memory. Other on-demand GPUs range from $0.99/hr (RTX 4090) to $7.10/hr (B300).
How Much Do Modal Sandboxes Cost With a GPU?
Modal Sandboxes use Modal's standard GPU rates, such as $3.95/hr for an H100, and are preemptible only. Sandbox CPU and memory cost about $0.071/vCPU-hr and $0.024/GiB-hr. Region pinning adds 1.15x or 1.75x.
Do GPU Sandboxes Charge for CPU and Memory on Top of the GPU?
Yes, on Thunder Compute, Daytona and Modal. vCPU and memory are billed per second alongside the GPU, adding $0.92-1.34/hr to an H100 sandbox with 8 vCPUs and 32 GiB of memory. Thunder Compute and Daytona also bill storage.
How Much Do 1,000 Eval Runs Cost on a GPU Sandbox?
1,000 eval runs of 20 minutes on one H100 with 8 vCPUs and 32 GiB cost about $1,292 on Thunder Compute, $1,625 on Daytona on-demand and $1,762 on Modal. Daytona preemptible costs $1,065 without evictions.
Is Spot or Preemptible Pricing Worth It for GPU Sandboxes?
Preemptible pricing pays off for short jobs that a harness can retry cheaply. For long evals and RL runs, rerunning evicted work can outweigh the discount, which makes a non-preemptible rate the safer choice.