Pricing

1. Serverless: pay per token

Call https://api.run.denv.ai/v1 with a key. It speaks the OpenAI API: POST /v1/chat/completions and GET /v1/models. You pay per token from a prepaid balance.

• Prices are per million tokens, counted by the engine for each request. Reasoning tokens are output tokens.
• New accounts get $5 of free credit to try it (GitHub accounts older than 30 days, once per person).
• Add credit by card, from $10 to $5,000 at a time, through Stripe. Turn on automatic top-up so you do not run out.
• Bigger top-ups get bonus credit: +10% on $100+, +20% on $500+, +30% on $2,000+. Pay $500, get $600 of credit.
• At zero balance the endpoint answers 402 until you add credit.
• Credit is not refundable, except where the law requires it.

Get a key and add credit in the console under Billing. Then follow Call the shared endpoint. What we keep and what we do not: data policy.

2. Through OpenRouter

We are applying to OpenRouter as a provider. Once listed, you can reach the same models through your OpenRouter account and pay OpenRouter at the prices it shows.

3. denv on your own cloud

denv deploys and runs models on machines in your own cloud account. You pay the cloud for the machines, as you do today.

• Free up to 8 GPUs running at once.
• Above 8 GPUs at once, 8% of the machine cost of that hour. An hour in which you never ran more than 8 GPUs costs nothing. Billed monthly.
• To run more than 8 GPUs at once, add a card.

For example, H100s at $4.50 an hour each, for one hour:

Start with the quickstart, or connect a GPU cloud.

Terms: terms of service.