Runcrate
On-demand GPUs and inference for AI teams and the agents they run, including GPU machines with real shells and an OpenAI-compatible model API.
Past recognitions
About Runcrate
On-demand GPUs and inference for AI teams and the agents they run, including GPU machines with real shells and an OpenAI-compatible model API.
Products
Named-GPU cloud instances and crates with a real shell for training, fine-tuning, serving, research, and agent jobs; billed for runtime.
An OpenAI-compatible API for hosted open and frontier models, including language, image, video, speech, and audio models, billed per use.
Persistent volumes that attach to GPU instances across regions, keep files after instance termination, and provide local-disk-speed access.
A cluster reservation service that matches multi-node GPU requirements with GPU-owner offers and delivers the selected cluster into the Runcrate dashboard.
Security and compliance
As reported by Runcrate. Ask for audit reports before you buy.
- GPU instances run in isolated containers inaccessible to other users.
- Workspace storage volumes are private and encrypted at rest.
- API keys are shown in full only once, stored securely, transmitted over HTTPS, and can be deactivated or deleted.
- Workspace roles provide Owner, Manager, and Developer access controls.
- Audit logs record significant workspace actions, including actor, timestamp, IP address, and changes.
- Payment processing is handled by Stripe, which is described as PCI DSS Level 1 certified; this is not evidence of Runcrate certification.
- Runcrate states that it uses encryption in transit and at rest, regular security audits, and strict access controls.
- Prompts and inference responses are not retained unless enabled or requested; saved playground and environment data have stated 30-day retention periods.
Is this your company?
Claim the profile to keep facts current, answer buyers and submit evidence for the next ranking review.
Claim Runcrate