API documentation.

grunden.ai is in production. Prices, models and features may still change, and breaking changes are published with notice. Planning something business-critical — get in touch at hej@grunden.ai so we can check first.

1. Create an account if you don't already have one.
2. Go to Settings → API keys in your team and create a key.
3. Top up your wallet under Wallet to pay per token, or activate a chat subscription under Subscription if you primarily use the web chat.

All requests go to https://api.grunden.ai. Send your key in the Authorization header:

Authorization: Bearer sk-grunden-…

Keys begin with sk-grunden- and are scoped per team. The rate limit per key (RPM + TPM) is set at creation and can be adjusted in the admin UI.

curl https://api.grunden.ai/v1/chat/completions \
  -H "Authorization: Bearer $GRUNDEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "glm-5.3-flash",
    "messages": [{"role": "user", "content": "Hej!"}]
  }'

Because the API is OpenAI-compatible, most AI coding tools can point at grunden.ai directly — swap out Claude/Codex as the backend without switching tools. Across the board: set the base URL to https://api.grunden.ai/v1, your sk-grunden-… key as the API key and glm-5.3-flash as the model. Calls run on the normal tier and are billed per token from your wallet (top up under Billing).

Se chat-kapaciteten →

We also expose an Anthropic-compatible POST /v1/messages endpoint, so Claude Code and other Anthropic-SDK-based tools can point here. You get our model GLM 5.3 Flash in Anthropic's format — we do not run Claude or other Anthropic models (EU-hosted, no US providers). Set in your shell:

export ANTHROPIC_BASE_URL=https://api.grunden.ai
export ANTHROPIC_API_KEY=sk-grunden-…

…and run claude as usual. Supported: text conversations, tool use, streaming. Not supported (yet): image blocks (replaced with a text placeholder), GLM's thinking mode isn't surfaced, prompt caching is ignored.