Skip to main content
Runcrate is built around two products on a single account:
  • Inference Engine: One OpenAI-compatible API for open and frontier models. Billed per token.
  • Compute: On-demand GPU instances, persistent storage, and dedicated clusters. Containers, VMs, or bare-metal.
Everything is managed from one dashboard with shared billing.
Runcrate is built for AI teams running real workloads:
  • AI product teams building inference-heavy features on open-source models
  • ML teams that need on-demand GPUs for training, fine-tuning, or evaluation
  • AI companies that have outgrown aggregators and want predictable per-token pricing or reserved capacity
  • Research labs that need bare-metal access without long-term commitments
Open and frontier models across 8 categories: Chat, Reasoning, Code, Vision, Image Generation, Video Generation, Text-to-Speech, and Speech-to-Text. Families include Llama, DeepSeek, Qwen, GLM, Kimi, Mistral, FLUX, and more. See the Model Catalog for the full list.
No. The Models API is accessible via HTTP requests from any language or framework. The dashboard is fully web-based. For GPU instances, you only need an SSH client (built into macOS, Linux, and Windows).
https://api.runcrate.ai/v1. The Models API is OpenAI-compatible: set it as base_url in the OpenAI Python SDK or baseURL in the JavaScript SDK, and use your Runcrate API key. The API keys page in the dashboard shows it with a copy button.
For inference on five open models, yes: sign in with GitHub for free usage, and no credit card is needed. A GitHub account that is too new waits first, and the API keys page shows the date free usage starts. If you sign up with Google, you need a card on file to use free models: free usage comes first, and anything beyond it is charged to your card by automatic top-up. Free usage is limited and resets on its own; your dashboard shows how much is left. After that they cost half the list price on shared capacity, and the list price when shared capacity is busy. See runcrate.ai/inference/free. Everything else uses prepaid credit: you add credit and only pay for what you use, starting from $20.
  • Models API: Instant. Make your first API call as soon as you have a key.
  • GPU Instances: Most instances are running within a few minutes.