On-demand GPU cloud compute for AI, LLM inference and development
European AI Gateway for accessing 700+ models through one API
Stage
Not given
Not given
Founded
Not given
2023
Based in
AE
SE
Pricing model
Paid
Freemium
Pricing
Not given
Free models available at signup with no credit card required. Gateway plan: 3% platform fee on credits. Control Plane plan: 5.5% platform fee on credits plus flat-rate per-check billing for LLM checks. Token rates are provider rates with no markup. No minimums, no lock-in. Enterprise plans include v
Key features
Live GPU marketplace with real-time capacity and pricing
Wallet-first authorization with fund reservation before compute request
SSH-ready GPU instances with public key access
Serverless inference with supported model catalog and API
Deploy templates for repeatable environments
Scoped developer API for capacity, usage, and inference automation
Support for open-source AI models on GPU instances
Multiple compute regions with dynamic availability
39 available GPU offerings with 7 GPU types
50 serverless models in model catalog
Access to 700+ models from 50+ providers
Spend limits and cost controls by project, role, user or key
Zero data retention by default with customizable trace storage
OpenAI SDK compatible with streaming and tool calling
Model routing with fallback options
LLM checks for flagging, redacting, blocking or scoring
Management API for infrastructure-as-code
SSO/SAML and role-based access control
JevMan Pac-Man benchmark for decision models
AI Roundtable for multi-model reasoning comparison
What makes it different
Server-side pricing calculation with transparent customer rates shown before authorization
Wallet-first deployment model prevents surprise charges
Clear operational state and control without opaque claims or sales processes
Flexible deployment options with persistent instances or serverless workflows
Developer-first API with scoped credentials and idempotency keys
No token markup: provider rates passed through unchanged
Zero data retention by default with flexible options
Hostnot GPU
Opper