Pricing

Start self-serve, then scale into dedicated capacity.

New accounts get $200 in free credit. Teams with production volume, private model requirements, or reserved capacity needs can work directly with us on a custom deployment.

Self-serve API

$200 free credit

Start building immediately with an OpenAI-compatible API key and usage-based inference.

Get API Key

Dedicated capacity

Custom

Reserve infrastructure for production workloads that need predictable capacity and support.

Contact Sales

Bring your model

Custom

Deploy private weights on General Compute infrastructure with the serving layer tuned for your workload.

Talk Pricing

What affects pricing

Workload shape: prompt length, output length, streaming, and concurrency all affect serving economics.

Capacity model: self-serve API usage is different from reserved infrastructure with deployment-specific guarantees.

Model ownership: hosted models and private weights have different operational requirements.