Same model, up to 4.66x different price — full Inference Providers pricing matrix

A new Hugging Face dataset from hardik90/hf-inference-pricing-matrix reveals that the same AI model can cost up to 4.66x more depending on the inference provider, based on an analysis of 14 providers, 107 models, and 295 listings from the Hugging Face router. The price spread for OpenAI's gpt-oss-120b across 10 providers is 4.41x, with DeepInfra often the cheapest, and the dataset is refreshed monthly to help users optimize inference costs.

Pulled every model × provider listing from the HF router /v1/models — 14 providers, 107 models, 295 listings. - openai/gpt-oss-120b: 10 providers, 4.41x price spread same weights, same outputs - Worst spread: Qwen3-235B-A22B 4.66x - deepinfra cheapest for most models; latency leaders vary Full matrix with cheapest/fastest provider per model: hardik90/hf-inference-pricing-matrix hardik90/hf-inference-pricing-matrix · Datasets at Hugging Face https://huggingface.co/datasets/hardik90/hf-inference-pricing-matrix If your inference bill matters, check your provider pin before your next month’s spend. Refreshed monthly.