Open-weight AI models, metered to the token.
COGS Moorhaven runs moorhaven-galleon-1 and moorhaven-skiff-1 behind one API, on our own servers in the United States. Join the waitlist; once you're in, create a key and pay only for the tokens you use. We never store your prompts or responses, and we never train on them.
curl https://moorhaven.ai/v1/chat/completions \ -H "Authorization: Bearer $MOORHAVEN_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "moorhaven-galleon-1", "messages": [{"role": "user", "content": "Plan the 9:30 ferry and a day on Nantucket"}] }'
OpenAI Chat Completions at /v1/chat/completions and Anthropic Messages at /v1/messages, so existing SDKs and tools work with a base-URL change. Responses can stream.
Pay for the tokens you use at the model rates below. The console shows usage by model and by member as it happens.
Invite members, and roles decide who can create API keys, change limits and manage billing. Rate limits are visible, not guessed.
Prompts and responses are processed in memory and never stored. Nothing you send is used to train AI models.
The Developer plan is free and starts with trial credit. Create a key in the console and call the API from any OpenAI or Anthropic SDK.
Join the waitlistMembers share keys, usage and billing under one organization. Roles decide who can create keys, change limits and manage billing.
Talk to usModels
Per-token rates apply on every plan, past any included usage. Model details and licenses.
| Model | Runs on | Input / 1M tokens | Output / 1M tokens |
|---|---|---|---|
| moorhaven-galleon-1 | Qwen3.8-27B (FP8) | $2.50 | $10.00 |
| moorhaven-skiff-1 | Qwen3.8-27B (FP8) | $0.40 | $1.60 |
Your data stays yours
We bill the tokens, not the content. Once a response is returned, the conversation exists only on your side. See exactly what we keep.
Requests are processed in memory and discarded once the response is returned. It applies to every organization on every plan, with no opt-in.
Your prompts, outputs and tool results are never used to train or improve AI models. Not anonymized, not aggregated: not used.
Every request travels over HTTPS. API keys and sign-in links are stored only as one-way hashes, and connector tokens are encrypted at rest.
Pricing
Usage is metered at the model rates above; Pro includes usage each month.
Try it free, no card needed.
- $5 of free trial credit
- API keys and the web console
- Usage dashboard
- Upgrade to Pro when the credit runs out
Includes $50 of usage per organization each month.
- Usage past the included amount at the model rates
- Invite your team, with roles and permissions
- Billing through Stripe, cancel any time
- Email support
For larger teams and special requirements.
- Custom rate and context limits
- Volume pricing
- Invoicing by arrangement
Prices in US dollars, before tax. See the Terms of Service for billing details.
Join the waitlist
Early access is opening in small groups. Leave your work email and we'll send you an invite when there's room.