Make your first API call.
Set up your key and model. Follow complete Python, JavaScript or curl examples, then add streaming and troubleshoot errors.
Get started with Models → PricingKnow what a request costs.
Read the billing unit and hourly multiplier. Walk through a worked example.
Understand dynamic pricing → API referenceFind the right endpoint.
Request parameters, response shapes and errors, with examples for every route.
Explore the API reference → AI Fabric · WaitlistRun your first GPU function.
Install the Python SDK, connect your machine and run a function on the fleet in four steps.
Get started with AI Fabric → AI Fabric · WaitlistBring your own model.
Prepare the runtime, model weights and workload requirements for onboarding.
Prepare a GPU deployment →API reference and model details
Models is Blade’s hosted inference API. Use https://api.models.blade.sh/v1 with your API key and an available model ID. Start with the guides and reference below.
- API routes, accepted parameters and errors
- Model catalogue
- Model pricing
- Hosting, external tools and data handling
- Model lifecycle: Preview, General Availability, End of Life
For agents and tools
Use the catalogue below for this site’s published model rates. The gateway documentation index covers the API protocol; some of those documents still use the legacy name Shadow Inference.
- Model catalogue (Markdown)
- Gateway documentation index (llms.txt)
The catalogue contains 70 models with From prices in EUR: the lowest price the dynamic multiplier can reach for each model. Check model access, supported parameters and the current rate in your console before integrating.
Running your own Python on GPUs?
Explore the AI Fabric SDK reference for functions, deployments, storage and autoscaling, or prepare your workload for onboarding.