Prices
Prices are in US dollars and exclude tax.
Serverless API
Available now.
Pay for the tokens you use. There is no minimum.
| Model | Input per 1M tokens | Output per 1M tokens |
|---|---|---|
| DeepSeek V4.1 Flash | $0.30 | $1.20 |
| Kimi K3.1 Pro | $2.20 | $11.00 |
Also served: Qwen, Hy4, embedding models, and voice and transcription models. Ask us for ids and prices.
Estimate a monthly bill
Batch
On request.
Send work that can wait and pay half the serverless price. Results come back within 24 hours.
Dedicated endpoint
On request.
One model on GPUs reserved for you, in one region. You are billed per GPU-hour, by the minute, whether or not requests are running.
| Term | Price per GPU-hour |
|---|---|
| On demand | $2.10 |
| One-month commitment | $1.85 |
Prices are per GPU on our standard GPU. Models that need larger GPUs are quoted.
Reserved node
On request.
Sovereign capacity for regulated institutions. A reserved node is hardware dedicated to you, in the country you choose, on a 12 to 36 month contract. You pay a fixed price for the capacity you reserve.
| Tier | What it is | From, per reserved GPU-hour |
|---|---|---|
| Standard | One site. Failed work reruns on your other nodes. | $1.70 |
Reserved nodes are delivered at Shoreheld sites. Tell us the country and the size and we will quote a delivery date.
Gateway and pilot
On request.
The gateway software is free and runs in your cloud account.
A four-week pilot runs open models on your own merged pull requests, beside the model you use today, and reports the cost per solved task. Pilots are quoted.
How the gateway worksStart with an API key
Token pricing needs no contract. Create a key and make a request.