Provider

Hugging Face Inference API free limit

Hugging Face Inference Providers is one API for open models from many inference providers. It used to include a small free monthly credit for every account. Today the official pricing page lists no monthly credit for free users. This guide explains the current rules, the 402 error and free options.

The current free limit

The Hugging Face pricing page for Inference Providers has a table of monthly credits per account type. Free users get no monthly credit. They can use Inference Providers only when they buy credits or upgrade to PRO.

Inference Providers monthly credits, from the Hugging Face pricing page (checked 20 November 2026)
Account typeMonthly creditPay as you go
Free userNoneYes, credit purchase required
PRO user$2.00Yes
Team or Enterprise organization$2.00 per seat, shared in the organizationYes
Older figure: many pages still say that free users get $0.10 per month. That figure is not on the official pricing page today. Third-party trackers recorded the change in October 2026. Older Hugging Face docs pages can still show the old value.

The PRO and Team credits are general compute credits. You can also spend them on Inference Endpoints, Spaces hardware and Jobs. Hugging Face charges the provider rate with no extra fee.

Why you get error 402

HTTP 402 means payment required. On Inference Providers, you usually get it when your account has no credit left. On a free account with no purchased credit, the first paid request can return 402.

CauseFix
Free account with no purchased creditBuy credits on the billing page, or upgrade to PRO
PRO credit for this month is spentWait for the next month, or buy more credits
The request bills an organization that has no creditRemove the X-HF-Bill-To header, or ask the organization admin to add credit

A different error can occur when an organization disables a provider. If you request that provider by name, the request fails. Use automatic provider selection or choose another provider.

Check your usage on the Inference Providers settings page. It shows usage for the past month by model and by provider.

How the API works

  1. Sign in to Hugging Face and open the token settings.
  2. Create a fine-grained token.
  3. Give the token the permission "Make calls to Inference Providers".
  4. Store the token in the HF_TOKEN environment variable.
  5. Buy credits or upgrade to PRO. Without credit, a free account cannot complete paid requests.

Inference Providers has an OpenAI-compatible endpoint for chat completions at https://router.huggingface.co/v1. You authenticate with a fine-grained Hugging Face token that has permission to make calls to Inference Providers.

OpenAI client with a Hugging Face token

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://router.huggingface.co/v1",
    api_key=os.environ["HF_TOKEN"],
)

resp = client.chat.completions.create(
    model="deepseek-ai/DeepSeek-V4.1-Flash:fastest",
    messages=[{"role": "user", "content": "Hello"}],
)
print(resp.choices[0].message.content)
  • Add :fastest to the model id to pick the provider with the highest throughput.
  • Add :cheapest to pick the provider with the lowest price.
  • Add a provider name, for example :novita, to pick one provider.
  • The OpenAI-compatible endpoint is for chat only. Use the Hugging Face client libraries for images, embeddings and speech.

What Inference Providers is good for

  • Testing many open models with one token, before you pick one for production.
  • Models that are rare elsewhere, for example new open-weight releases and special task models.
  • Image, video and audio models through the Hugging Face client libraries.
  • Teams that already pay for PRO or Team, because the included credit applies first.

You can also download many open models from the Hub and run them on your own hardware. Tools such as llama.cpp, vLLM and Ollama serve them locally. Local inference costs no API credit, but you need a GPU or a fast computer. FreeLLMAPI can add a local OpenAI-compatible server as a custom provider, next to the free cloud tiers.

Inference Providers is not a free API for daily use. With no monthly credit, a free account pays for every request after the purchase. The $2.00 PRO credit is also small for an agent or a chat app.

Free options for open models

Many open models on Hugging Face also have recurring free tiers at other providers. The table shows the strongest free models in today's catalog.

Strongest free models in the catalog today
ModelFree onContextFree limits (best offer)ToolsVision
GLM 5.3APINex, UnoRouter1M5 req/min
10M weighted tokens/day shared across APINex free routes; 4x token multiplier; 5 requests/minute/account
YesNo
Qwen3.8 MaxACLIDE-€20/month shared internal usage credits · no cardYesNo
GLM 5.3 FlashAPINex, BlazeAPI, NVIDIA NIM +1 more1M40 req/min
40 RPM
YesYes
Gemini 3.8 FlashAPINex, Google AI Studio1M5 req/min
10M weighted tokens/day shared across APINex free routes; 3x token multiplier; 5 requests/minute/account
YesYes
Gemini Flash (latest alias)Google AI Studio-10 req/min
20 req/day
250,000 tokens/min
~3M
YesYes
Qwen3.8 Flash NextAMD Radeon Cloud, BlazeAPI, ElectronHub262KRecurring daily account allowanceYesNo
Qwen3.8 FlashElectronHub, Inferbase1M$0.25/week shared recurring creditYesYes
Deepseek V4.1 FlashAPINex1M5 req/min
10M weighted tokens/day shared across APINex free routes; 3x token multiplier; 5 requests/minute/account
NoNo
Gemini 3.7 FlashGoogle AI Studio1M10 req/min
20 req/day
250,000 tokens/min
~3M
YesYes
GPT-5.6 LunaElectronHub1M$0.25/week shared recurring creditYesYes
Dots3 Note PreviewKilo Gateway, OpenRouter512K20 req/min
50 req/day
50 req/day account-wide (1,000 after a $10 purchase)
YesYes
Agnes 3.0 FlashAgnes AI-Promo $0/tokenNoNo

Showing 12 of 354 matching free models. See all models.

Live data from the FreeLLMAPI catalog (version 2026.11.19, checked 19 November 2026). Free limits change often. The catalog is checked every day.

FreeLLMAPI has a Hugging Face adapter. Today's free catalog has no Hugging Face rows, because free accounts have no recurring credit. To use free open models with one key, add keys from providers in the table. The router moves to the next provider when one returns a rate limit error.

Use every one of these through one key. FreeLLMAPI is an open-source, self-hosted router that puts all these free tiers behind a single OpenAI-compatible endpoint and fails over when one is rate-limited. Browse the catalog or go live.

Frequently asked questions

Is the Hugging Face Inference API free?

Not for free accounts. The current pricing page lists no monthly credit for free users. They must buy credits or upgrade to PRO, which includes $2.00 of compute credit per month.

Do free Hugging Face users still get $0.10 per month?

The official pricing page does not show it today. It lists no monthly credit for free users. Older pages and some docs mirrors still show the $0.10 figure.

Why does Hugging Face return error 402 Payment Required?

Your account or the billed organization has no credit left. Buy credits, upgrade to PRO, or wait for the next monthly PRO credit.

Is the Hugging Face router OpenAI-compatible?

Yes, for chat completions. Set the base URL to https://router.huggingface.co/v1 and use a Hugging Face token as the API key.

What is a free alternative to Hugging Face Inference?

Use providers with recurring free tiers for open models, such as Groq, NVIDIA NIM, Cloudflare Workers AI or OVH AI Endpoints. FreeLLMAPI pools them behind one key.

Sources

All guides · Browse all free models

Keep reading

All posts →
Provider

Free Gemini API

Google AI Studio's permanent free tier — Gemini and Gemma, no credit card.

Provider

Free DeepSeek API

DeepSeek V3, R1 and V4 served free across multiple providers.

Provider

Free Llama API

Meta's Llama models free across 14 providers, behind one key.