Hugging Face Inference Providers

permanent free · aggregator · verification: partially_verified (75%, doc_fetch, 2026-07-17) · setup guide →

The record

Base URLs	https://router.huggingface.co/v1 openai-compatible
API key required	yes
Credit card	no
Phone verification	no
May train on prompts	unknown
Quota	free_tier [monthly_credits] “Free Users: $0.10 monthly credits, "subject to change"; PRO Users: $2.00; Team or Enterprise Organizations: $2.00 per seat. "Every Hugging Face user receives monthly credits to experiment with Inference Providers."”
Signup friction	3 steps · ~5m to first key
Env var	HF_TOKEN
Notes	Router base URL confirmed from official docs; described as "drop-in replacement for the OpenAI chat completions API" — chat completion tasks only (openai_responses false; other tasks need HF InferenceClient). Provider selection via model suffix: :fastest (default), :cheapest, :preferred, or :<provider-name>. Automatic failover across providers with provider="auto".

model	capabilities	context	free	limits
Qwen/Qwen2.5-7B-Instruct	chat	131,072	yes	"Credit-metered" (mnfst); $0.10/month free credits shared across all models and providers
deepseek-ai/DeepSeek-V3-0324	chat	?	yes	Credit-metered: "$0.10, subject to change" monthly credits for free users, shared across all models and providers
meta-llama/Meta-Llama-3.1-8B-Instruct	chat	128,000	yes	"Credit-metered" (mnfst); $0.10/month free credits shared across all models and providers
openai/gpt-oss-120b	chat, reasoning, tool_calling	?	yes	Credit-metered: "$0.10, subject to change" monthly credits for free users, shared across all models and providers