In one line
Usage-based inference across model types, with dedicated deployments and selected external model APIs.
- Paid
- 4 sources
What is DeepInfra?
DeepInfra offers language, image, video, speech, embedding and reranking APIs for public and custom models. Locally hosted inference and models forwarded to external providers have different data flows. Published policy limits content retention for DeepInfra-hosted inference, with exceptions for image outputs, batch, paid caching and authorized troubleshooting. External Google or Anthropic models forward data under upstream policies; operational metadata can remain.
What can you use DeepInfra for?
- Access supported models through an API
- Manage model usage and capacity
- Review data handling and deployment options
DeepInfra pricing and usage limits
Pricing varies by model: tokens, inference execution time or media units. Dedicated models are billed for deployed time. A payment method or prepaid balance is required; consult each model's price.
This site never quotes prices. Confirm amounts on the official site.
Sources checked
These pages were opened during verification. Details may have changed; the official pages win.
Sources last checked for this entry:
This date records a check of this entry against the linked sources. It does not mean hands-on testing, continuous monitoring, or a security audit. How sources are checked
FAQ
Why does every API relay show a disclaimer?
A relay handles keys, data, and charges. This directory has not audited those sites. The fixed line is: Unverified, not an endorsement. Assess the risk yourself; you are responsible for your own data and funds.
Can an API relay be recommended or sponsored?
No. An API relay cannot be featured or sponsored and never appears in a comparison, and the relay category has no sponsor slot.