Connect Hugging Face to StackSpend
Track Hugging Face costs including Inference Endpoints, Spaces (Gradio), Jobs, Inference Providers, and Private Storage.
StackSpend tracks Hugging Face costs for Inference Endpoints, Spaces, Jobs, and Storage. Get daily visibility, unified view with OpenAI and Anthropic, and anomaly alerts. Open-source and closed-source AI costs in one place.
Organization Account Required
Hugging Face billing data is only available for organization accounts. You need organization owner or admin access to create a token with billing permissions.
Prerequisites
- A Hugging Face organization (Team or Enterprise plan recommended)
- Organization owner or admin access
- Your organization slug (e.g. from huggingface.co/your-org)
Create an Access Token
Create a token at Hugging Face with billing access. You can use either Fine-grained (recommended) or Read token.
- 1Go to Hugging Face Settings → Access Tokens
- 2Click Create new token
- 3Fine-grained (recommended): Under User permissions, enable "Read access to your billing usage and know if a payment method is set". Under Org permissions, select your organization and enable "Read access to organizations settings". Together these grant billing access for that org.
- 4Or use a Read token: Simpler setup with broader access. If you have org admin access, a Read token will work.
- 5Copy the token immediately (starts with
hf_)
Direct link for Fine-grained token with billing pre-selected: Create token with billing
Get Your Organization Slug
Your organization slug is in the URL when you view your org on Hugging Face.
Go to your organization's page (e.g. huggingface.co/your-org) and copy the org name from the URL.
Connect in StackSpend
Add Hugging Face as a provider in StackSpend and enter your credentials.
- 1In StackSpend, go to Providers and click Add Provider
- 2Select Hugging Face
- 3Paste your Access Token and enter your Organization slug
- 4Click Test Connection to verify, then Save
- 5Click Sync to fetch your billing data
What Gets Tracked
- Inference Endpoints — GPU/CPU compute costs by endpoint and model
- Spaces — Gradio Spaces compute usage
- Jobs — CPU/GPU job costs
- Inference Providers — Serverless inference usage
- Private Storage — Storage above included limits