About DeepInfra
DeepInfra hosts open AI models behind an API and bills usage by the token. It also rents GPUs for teams that run their own workloads.
What it does
Developers pick a model from the catalog and call it with an API key. The catalog includes text models from many open-model teams, and some listings show a discount for a set period. GPU instances and clusters are offered separately for workloads that need dedicated hardware.
Key features
- Hosted text models billed per token
- Catalog from many open-model makers
- Simple developer API access
- On-demand GPU instances for rent
- Dedicated clusters arranged with sales
Who it's for
- Startups that want to keep model costs low
- Developers running several open models side by side
- Teams that want to move from API use to their own GPUs
Pricing and good to know
Prices are listed per million input and output tokens and differ by model. Some models are shown with a percentage discount. A consultation can be booked for larger deployments. Model versions and prices change often, so check the catalog before you finalize a budget.
Highlights
- Call hosted text models billed per token
- Choose from a catalog of open models
- Access models through a simple API
- Rent on-demand GPU instances
- Arrange dedicated clusters with sales
