Introduction
Helmcode is private AI inference for European teams. Connect your tools to open models running on EU infrastructure through an OpenAI-compatible API — swap the base URL and everything keeps working unchanged.
- OpenAI-compatible. Every OpenAI SDK and tool works as-is — Cursor, Zed, OpenCode, your own clients.
- Unlimited tokens on Qwen and Gemma. One flat rate per API key, not per token, for every model running on our own infrastructure. DeepSeek V4 Flash and GLM 5.3 Flash share a monthly quota set by your plan, and every key is limited to 2M tokens per minute: see Rate limits. The frontier models we resell from Anthropic, OpenAI and Google sit outside the flat rate and are paid per token from prepaid credit.
- Zero logs, EU data. Prompts are never stored, and every model we run runs on EU infrastructure, never on US hyperscalers subject to the Cloud Act. Leaving the EU is a choice you make model by model: pick one of the frontier models we resell and that request is sent to whichever of Anthropic, OpenAI or Google serves the model you picked, processed on that provider’s infrastructure outside the EU and under their terms. Every request on any other model stays on hardware we operate in the EU.
How it works
Your team manages API keys from the Helmcode Dashboard. Each key authenticates requests to https://api.helmcode.com/v1 and carries its own rate limits — RPM, concurrency and 2M tokens per minute.
Helmcode is built for teams and organizations. API keys are issued per workspace and scoped by your plan — not to individual people. Manage members, keys, and usage from one place.
Base URL
https://api.helmcode.com/v1Next steps
- Quickstart: create a key and make your first request.
- Models: the full catalog and model IDs.
- Examples: code samples in Python, Node.js, and cURL.
- Set up your agent: point Claude Code, Cursor, Codex, VS Code, OpenCode, or Zed at Helmcode.