Home / AI agent sandboxes / Tensorlake Sandboxes / Pricing
Tensorlake Sandboxes pricing
In short (2026-11-14): $0.07 per vCPU-hour of active CPU, $0.015 per GiB-hour of memory, plans from $250.00 a month. Free tier: Free plan at $0 with no credit card: 1 sandbox at a time with 1 vCPU, 1 GB RAM and 10 GB disk, sessions up to 2 hours, unmetered. Source: official pricing page. Prices checked 2026-11-15.
Usage rates
- vCPU
- $0.07 per vCPU-hour, billed only while the CPU is active
- Memory
- $0.015 per GiB-hour
- Storage
- $0.07 per GiB-month
- Billing
- per second, billed in whole compute units ($0.01 each), pooled per organization. CPU: active execution time, with a minimum of 20% of allocated vCPU (full allocation when active metering is unavailable). RAM and root disk: allocated amount while running. Suspended sandboxes pay snapshot storage only.
Plans
| Plan | Per month | Included credit | Limits |
|---|---|---|---|
| Free | $0.00 | $0.00 | 1 sandbox at a time, 1 vCPU, 1 GB RAM, 10 GB disk, sessions up to 2 hours. Sessions are not metered; no paid usage is possible on this plan. |
| Usage Credits | $0.00 | $0.00 | Prepaid packs of $5, $10 or $20 (500-2,000 compute units); no postpaid overage, sandboxes stop when the balance hits zero unless auto-refill is on. Up to 100 concurrent sandboxes, 4 vCPU, 16 GB RAM, 100 GB disk, sessions up to 24 hours. |
| Pro | $250.00 | $250.00 | $250/month including 25,000 compute units ($250) of usage at Pro rates, then $0.01 per unit. 1,000 concurrent sandboxes, up to 16 vCPU, 64 GB RAM, 100 GB disk, no session time limit. HIPAA with a BAA. |
GPU prices
- NVIDIA H100 (reserved capacity, separate invoice): $3.95 per hour
What it costs in practice
| Workload | Usage | Cheapest plan | Total per month |
|---|---|---|---|
| Hobby agent: 1 vCPU, 1 GiB, 20 hours a month | $0.72 | Usage Credits | $0.72 |
| One dev agent: 2 vCPU, 4 GiB, 8 h a day for 22 days | $17.95 | Usage Credits | $17.95 |
| Team: 10 sandboxes of 2 vCPU, 4 GiB, 8 h a day for 22 days | $179.52 | Usage Credits | $179.52 |
| Always on: 4 vCPU, 8 GiB, 730 hours a month | $189.80 | Usage Credits | $189.80 |
CPU busy 30% (50% always on), which matters only where CPU is billed by active time. Storage, network and GPU are not included. Try your own numbers in the cost calculator.
Billing details that change the bill
- Rates shown are the Usage Credits (pay-as-you-go) rates: $0.07 per active CPU core-hour and $0.015 per GB-hour of RAM. Pro rates are 40% lower: $0.042 per active core-hour and $0.009 per GB-hour of RAM, so the site's Pro estimate, which applies the higher rates against Pro's $250 credit, overstates Pro cost.
- Root disk is billed on allocated size while the sandbox runs: $0.0002 per GB-hour on Usage Credits, $0.0001 on Pro (about $0.146 and $0.073 per GB-month if run all month). The storage rate shown is snapshot storage, $0.07 per GB-month on both plans, billed while a sandbox is suspended. Network egress is free.
- Prices are quoted per GB; the usage tab in the dashboard reports GiB-hours. We treat GB as GiB.
- Memory must be between 1 GB and 8 GB per vCPU. Max size and session length depend on the plan: Free 1 vCPU / 1 GB / 2 hours, Usage Credits 4 vCPU / 16 GB / 24 hours, Pro 16 vCPU / 64 GB / no limit. The max size and session length shown are Pro's.
- Only named sandboxes can be suspended and resumed; ephemeral sandboxes run until terminated or timed out. A terminated sandbox can be restarted within 48 hours.
- CPU is billed on active execution but never below 20% of the allocated vCPU: an idle 8-vCPU sandbox bills 1.6 vCPU. The site's estimate does not apply this floor, so it understates cost for sandboxes whose CPU is busy less than 20% of the time.
- Conflict: the pricing page states the 20% CPU minimum; the billing docs page does not mention it and says only that CPU tracks active execution.
- GPU sandboxes are available by request, run on gVisor instead of VMs, and cannot be live resized. The pricing page lists reserved NVIDIA H100 capacity at $3.95 per GPU-hour, billed on its own invoice separate from sandbox usage; reservations go through sales.
- Running in your own cloud account (in-VPC or on-prem) is an Enterprise option with custom pricing.
- When a Usage Credits balance reaches zero, new sandbox activity is blocked and running sandboxes are stopped.
Price history
List price per month over timeNo change since 2026-11-13
No price change found since we started tracking on 2026-11-13. We check every pricing page daily.
Cheaper or better-fit Tensorlake Sandboxes alternatives →
Sources
- https://tensorlake.ai/pricing, checked 2026-11-14:
An idle 8-vCPU sandbox bills 1.6 vCPU; a busy one bills what it uses.
- https://tensorlake.ai/pricing, checked 2026-11-14:
Includes 25,000 CU ($250 of usage) per billing cycle. Additional usage is billed at $0.01 per CU at the Pro rates below.
- https://tensorlake.ai/pricing, checked 2026-11-14:
We block new sandbox activity and stop running sandboxes instead of turning the shortfall into an unexpected bill.
- https://tensorlake.ai/pricing, checked 2026-11-14:
The NVIDIA H100 is $3.95 per GPU-hour.
- https://docs.tensorlake.ai/platform/billing, checked 2026-11-14:
Active CPU per core-hour | $0.07 | $0.042 | Custom
- https://docs.tensorlake.ai/platform/billing, checked 2026-11-14:
RAM per GB-hour | $0.015 | $0.009 | Custom
- https://docs.tensorlake.ai/platform/billing, checked 2026-11-14:
Root disk per GB-hour | $0.0002 | $0.0001 | Custom
- https://docs.tensorlake.ai/platform/billing, checked 2026-11-14:
Snapshot storage per GB-month | $0.07 | $0.07 | Custom
- https://docs.tensorlake.ai/platform/billing, checked 2026-11-14:
Rates are quoted per hour for readability, but runtime usage is measured by the second.
- https://docs.tensorlake.ai/platform/billing, checked 2026-11-14:
Pro | $250 / month + usage | 1,000 | 16 vCPU, 64 GB RAM, 100 GB root disk | Unlimited
- https://docs.tensorlake.ai/platform/billing, checked 2026-11-14:
Free | $0 forever | 1 | 1 vCPU, 1 GB RAM, 10 GB root disk | 2 hours
- https://docs.tensorlake.ai/platform/billing, checked 2026-11-14:
allocate between 1 GB and 8 GB of RAM per vCPU
- https://docs.tensorlake.ai/sandboxes/introduction, checked 2026-11-14:
GPU sandboxes use gVisor while CPU-only sandboxes run in VMs.
- https://docs.tensorlake.ai/sandboxes/introduction, checked 2026-11-14:
Tensorlake is HIPAA and SOC 2 Type II compliant, supports EU data residency, and offers zero data retention.
- https://docs.tensorlake.ai/sandboxes/lifecycle, checked 2026-11-14:
Named sandbox is paused. Consumes no compute; state is preserved for resume under the same sandbox ID.
- https://docs.tensorlake.ai/sandboxes/gpu, checked 2026-11-14:
GPU sandboxes are available by request.
- https://docs.tensorlake.ai/sandboxes/networking, checked 2026-11-14:
Restricting the sandbox's own outbound internet access
- https://tensorlake.ai, checked 2026-11-14:
Each tool call and harness runs in its own microVM. No shared kernel, no cross-tenant state.