Prime Intellect — Full-stack platform for training custom AI agents: GPU compute marketplace, open-source RL framework (prime-rl, verifiers), environments and evals.. Pricing model: Usage-based: per GPU-hour compute marketplace + paid hosted tools; open-source frameworks free. Value metric: GPU-hours (chip type, quantity, duration).
At a glance
| Company | Prime Intellect |
|---|---|
| What it does | Full-stack platform for training custom AI agents: GPU compute marketplace, open-source RL framework (prime-rl, verifiers), environments and evals. |
| Headquarters | San Francisco, USA (ARR Club lists US) |
| Pricing model | Usage-based: per GPU-hour compute marketplace + paid hosted tools; open-source frameworks free |
| Value metric | GPU-hours (chip type, quantity, duration) |
| Pricing page | www.primeintellect.ai/pricing |
Plans and prices
Pay per GPU-hour, e.g. H100 from $1.65/hr, A100 from $0.87/hr (July 2024 launch post); modular hosted tools, no public tiers. Pricing page not fetchable.
Growth milestones
Prime Intellect appears in our analysis of 2026's $100M ARR class. Figures are company-reported unless stated.
- 2026-07-08: $130M Series A at $1B valuation led by Radical Ventures (Nvidia, Intel Capital, Dell, Iconiq); $100M annualised run rate; customers incl. Ramp, Zapier. (source)
- 2026-07-09: Surpassed $100M ARR in 12 months (ARR Club). (source)
How it grows
Open-source research releases (models, RL libraries) build developer community and credibility; modular marketplace lets customers adopt one piece at a time; enterprise 'own your AI' positioning.
What you can steal
low-budget growth
Open-source the framework and charge for hosting/compute: free tools become your lead generation.
land and expand
Let customers buy one module (compute) and add others (training, evals) later instead of forcing an all-in-one bundle.
Sources

Get both books free
The Subscription Playbook (ebook + audiobook) and Feed a Starving Crowd, instantly. In Australia? Claim the free hardcover.
