5 models from 4 makers that you can call through Hugging Face Inference — each one tested by us with fixed prompts, outputs kept, costs measured.
Makers: black-forest-labs · qwen · stabilityai · tongyi-mai
Docs Get an API key Their pricing page
Hugging Face is the largest open-weight model hub — over 100,000 models, most of which can be called through its Inference API without standing up your own infrastructure.
A free tier exists for light usage; dedicated Inference Endpoints are the path to production-grade serving. For a model with no other hosted API, Hugging Face is often the only way to call it without self-hosting.
Affiliate program: None found.
This is what Hugging Face Inference itself charges for each model — not the cheapest price across every provider (that's what the cards above show).
| Model | Maker | Price via Hugging Face Inference | Context |
|---|---|---|---|
| black-forest-labs/FLUX.1-dev | black-forest-labs | — | — |
| Qwen Image | qwen | — | — |
| stabilityai/stable-diffusion-3.5-medium | stabilityai | — | — |
| SD XL | stabilityai | — | — |
| Tongyi-MAI/Z-Image-Turbo | tongyi-mai | — | — |
Latency: Per-provider latency isn't independently measured yet — the probe rig currently tests one route per model, not one per provider (see PROBE_RIG.md ledger #10).