Prime Intellect Sandboxes
Prime Intellect runs each sandbox in a hardware-isolated microVM, with pause and resume that keeps memory and processes, checkpoints you can fork, and thousands of sandboxes starting at once. Built for RL rollouts, evals and agent fleets.
- 2 vCPU + 4 GiB
- $0.090/hr 0.54x E2B
- Isolation
- microVM
- Pause / resume
- Yes
- Max session
- 60 min default, no lifetime limit when set
Visit Prime Intellect Sandboxes Compare side by side Request a quote Checked
Before you buy
Prime Intellect’s $0.02 per vCPU-hour is a launch rate, valid through 22 December 2026. Source, seen 9 Nov 2026.
Priced next to Prime Intellect Sandboxes
All providers against E2B| Provider | 2 vCPU + 4 GiB$/hr | vs E2B | Relative cost | Isolation |
|---|---|---|---|---|
| Alibaba Cloud Agent Sandbox | $0.055 (source: Alibaba Cloud Agent Sandbox billing overview) | 0.33x | VM | |
| Northflank Sandboxes | $0.067 (source: Northflank Pricing) | 0.40x | Cloud Hypervisor microVM | |
| Prime Intellect Sandboxes | $0.090 (source: Prime Intellect Sandboxes docs) | 0.54x | microVM | |
| Freestyle | $0.132 (source: Freestyle Pricing) | 0.80x | VM | |
| Beam | $0.140 (source: Beam Pricing) | 0.85x | — |
Every field
Source on each value- Isolation Isolation technology the vendor states for each sandbox.
- microVM (source: Prime Intellect Sandboxes)
- Other isolation options Alternative runtimes the vendor offers besides the default (e.g. VM runtime, gVisor for GPU).
- Not published
- Deployment Managed cloud service, self-hosted software, or both.
- managed (source: Prime Intellect Sandboxes)
- Billing basis What the meter counts: wall-clock allocated resources, active CPU only, requests/plan quota, session-hours, or free software.
- wall clock (source: Prime Intellect Sandboxes docs)
- vCPU price $/vCPU-hr Published on-demand CPU rate normalised to USD per vCPU-hour (per-second rates multiplied by 3,600; per-core rates divided by vCPUs per core where the vendor defines it). Lowest public tier, default region.
- $0.02 (source: Prime Intellect Sandboxes docs)
- Memory price $/GiB-hr Published memory rate normalised to USD per GiB-hour (GB treated as GiB where the vendor does not distinguish). 0 means the vendor states memory is free.
- $0.0125 (source: Prime Intellect Sandboxes docs)
- Sandbox from $/sandbox-hr Smallest published per-sandbox hourly price, where the vendor publishes one.
- $0.0335 (source: Prime Intellect Sandboxes)
- Paid plan from $/mo Cheapest paid platform plan fee (before usage).
- Not published
- Free tier A free plan, free monthly allowance or sign-up credit exists.
- Not published
- Free tier detail What the free tier or credit includes, as published.
- Not published
- Cold start ms Vendor-published create/boot time in milliseconds. Only filled when the vendor publishes a number; see startup_claim for context (percentile, conditions).
- Not published
- Resume time ms Vendor-published resume-from-pause/standby/snapshot time in milliseconds.
- Not published
- Startup claim The vendor's startup/resume wording with conditions, quoted or closely paraphrased.
- "Thousands start within seconds" (no per-sandbox figure) (source: Prime Intellect Sandboxes)
- Max session length Longest single sandbox session by plan, as published.
- 60 min default, no lifetime limit when set (source: Prime Intellect Sandboxes docs)
- Persistence / snapshots Filesystem state can be kept between runs (snapshots, persistent disks, volumes).
- Yes (source: Prime Intellect Sandboxes docs)
- Pause / resume A sandbox can be paused, suspended, hibernated or put in standby and resumed later.
- Yes (source: Prime Intellect Sandboxes docs)
- Memory state kept Pause/hibernate preserves running memory state, not only the filesystem.
- Yes (source: Prime Intellect Sandboxes docs)
- Fork / branch A running or snapshotted sandbox can be cloned into new sandboxes.
- Yes (source: Prime Intellect Sandboxes docs)
- GPU GPU sandboxes are offered.
- No (source: Prime Intellect Sandboxes docs)
- Self-host Can run on your own hardware or cluster (open source or licensed).
- Not published
- BYOC / VPC Vendor-managed deployment inside your own cloud account.
- Not published
- Open source Core sandbox runtime is published under an open-source licence.
- Not published
- Licence Open-source licence of the runtime, if any.
- Not published
- SDKs / languages Official SDK languages (or supported execution languages where noted).
- Python, CLI (prime) (source: Prime Intellect Sandboxes docs)
- Regions Published regions or region count.
- Not published
- Max vCPU per sandbox vCPU Largest published vCPU count per sandbox on any self-serve or enterprise plan.
- 16 (source: Prime Intellect Sandboxes docs)
- Max memory per sandbox GiB Largest published memory per sandbox.
- 64 (source: Prime Intellect Sandboxes docs)
- Concurrency Published concurrent sandbox limits by plan.
- 1,024 active sandboxes (default account limit) (source: Prime Intellect Sandboxes docs)
- SOC 2 SOC 2 status as stated by the vendor.
- Not published
- HIPAA Vendor offers HIPAA-compliant usage/BAA (on any plan).
- Not published
- Pricing from
- $0.02 per vCPU-hour
- Pricing as published
- Launch rates valid through 22 December 2026: $0.02/vCPU-h, $0.0125/GiB-h, disk $0.0002/GiB-h, billed while running; a paused sandbox pays for disk only. 1 vCPU / 1 GiB / 5 GiB disk = $0.0335/h. Images and checkpoints $0.10/GiB-month. GPUs "coming soon". Pricing checked 2026-11-09.
Sources
- Prime Intellect Sandboxes accessed 2026-11-09
- Prime Intellect Sandboxes docs accessed 2026-11-09
Get pricing from Prime Intellect Sandboxes and two close alternatives
Tell us what you need once. We send it to matching providers and they reply to you directly.