Hugging Face pricing snapshot
Hugging Face charges in two layers. A plan sets account limits and organization controls: PRO for one person, Team and Enterprise per user for an organization. Compute and storage are priced separately by usage: Spaces hardware and Inference Endpoints by the hour, storage by the TB per month.
Scroll sideways to see all columns. With a keyboard, focus this table and use the arrow keys.
| Plan | Published price | What it adds |
|---|---|---|
Free | Free | Free CPU Basic Spaces hardware (2 vCPU, 16 GB) and the Free ZeroGPU quota; storage, ZeroGPU and API limits are in the plan details below |
PRO | $9/month | Personal account: hosting your own ZeroGPU, Gradio & Docker Spaces on compute, 10× private and 2× public storage capacity, $2 of included Inference Providers credits monthly, 8× ZeroGPU quota with highest queue priority, Spaces Dev Mode and the Dataset Viewer for private datasets |
Team | $20/month per user | Organization plan with Single Sign-On, audit logs, resource groups, token management, repository analytics and advanced security controls (enforce 2FA, default repo visibility, domain auto-join); paid self-serve by credit card or AWS Marketplace |
Enterprise | $50/month per user on the pricing page (Talk to sales); the Enterprise page shows Custom pricing | User provisioning (SCIM); token management adds revocation and service accounts; billed by invoice with an annual commitment; sales offers flexible contract options |
Choose a plan by what you need
One person who needs more quota. PRO is a personal upgrade. Pick it for storage, ZeroGPU quota or the monthly inference credits, not for organization controls.
An organization that needs access control. Team is the self-serve entry point for SSO, audit logs and resource groups with granular access control to repositories.
Procurement by invoice or larger included limits. Enterprise is the plan for invoice billing and the higher public storage, ZeroGPU and API limits shown below.
Priority support. Hugging Face lists priority support as available with Enterprise Plus, for the most advanced organizations.
Included limits by plan
Show full plan details
Scroll sideways to see all columns. With a keyboard, focus this table and use the arrow keys.
| Limit | Free | Team | Enterprise |
|---|---|---|---|
Private storage included | 100 GB | 1 TB per seat | 1 TB per seat |
Public storage included | Best-effort | 12 TB + 1 TB per seat | 200 TB + 1 TB per seat |
ZeroGPU quota per member | 5 min per day | 40 min per day | 60 min per day |
Hub API rate limit | 1,000 req / 5 min | 3,000 req / 5 min | 6,000 req / 5 min |
- Vendor source · Observed · Stored official documentation capture for hugging-face.
- Vendor source · Observed · Stored official documentation capture for hugging-face.
How compute is billed
Upgraded Spaces hardware has an hourly price and a monthly price for each option. The calculator below uses the hourly Spaces hardware prices for six options: enter how many Spaces run on each and the hours they run in a month.
Scroll sideways to see all columns. With a keyboard, focus this table and use the arrow keys.
| Spaces hardware | Hourly price | Monthly price |
|---|---|---|
CPU Upgrade (8 vCPU, 32 GB) | $0.03 | $22 |
Nvidia T4 - small (16 GB VRAM) | $0.40 | $288 |
1x Nvidia L4 (24 GB VRAM) | $0.80 | $576 |
Nvidia A10G - small (24 GB VRAM) | $1.00 | $720 |
1x Nvidia L40S (48 GB VRAM) | $1.80 | $1,296 |
Nvidia A100 - large (80 GB VRAM) | $2.50 | $1,800 |
Hugging Face Spaces cost calculator
Estimate monthly Spaces hardware costs from the hourly prices in Hugging Face's Spaces Hardware table: each option's price for every hour a Space runs on it.
Modeled scope: Hugging Face Spaces hardware · hourly prices · on demand
Example workload
Selected services
Checking current pricing…
Free CPU Basic and ZeroGPU hardware, other Spaces hardware options, Inference Endpoints and Inference Providers, storage, PRO, Team and Enterprise plans, and taxes are excluded.
Estimate based on public list pricing. This is not an official vendor quote.
Published pricing reference
These published rates are the inputs to the calculator. Taxes and negotiated discounts are excluded.
| Component | Rate and unit | Commitment and region | Verified | Source |
|---|---|---|---|---|
| CPU Upgrade | $0.03 USD per 1 instance-hour | Spaces hardware · On demand | Nov 9, 2026 | Official source |
| Nvidia T4 - small | $0.40 USD per 1 instance-hour | Spaces hardware · On demand | Nov 9, 2026 | Official source |
| 1x Nvidia L4 | $0.80 USD per 1 instance-hour | Spaces hardware · On demand | Nov 9, 2026 | Official source |
| Nvidia A10G - small | $1 USD per 1 instance-hour | Spaces hardware · On demand | Nov 9, 2026 | Official source |
| 1x Nvidia L40S | $1.80 USD per 1 instance-hour | Spaces hardware · On demand | Nov 9, 2026 | Official source |
| Nvidia A100 - large | $2.50 USD per 1 instance-hour | Spaces hardware · On demand | Nov 9, 2026 | Official source |
Estimate assumptions
Hugging Face Spaces hardware · hourly prices · on demand
View assumptions and exclusions
- Each Space runs on one hardware option for the hours entered; 730 hours is a Space that runs all month.
- Each option is charged at the hourly price in Hugging Face's Spaces Hardware table.
Free CPU Basic and ZeroGPU hardware, other Spaces hardware options, Inference Endpoints and Inference Providers, storage, PRO, Team and Enterprise plans, and taxes are excluded.
Other costs that move the bill
Inference Endpoints. Dedicated inference starts at $0.033/hour. GPU instances on AWS are priced per hour by GPU type and count, for example 1x NVIDIA T4 at $0.50, 1x A100 at $2.50 and 1x H100 at $4.50. CPU instances are also offered on Azure and GCP.
Inference Providers. Hugging Face describes access to 45,000+ models from leading AI providers through a single, unified API with no service fees. Team and Enterprise can enable organization billing with usage analytics and spending limits.
Storage beyond the included quota. Private storage past the plan quota is billed at $18 per TB per month, with volume discounts. Egress and CDN are included at no extra cost.
Side projects. Hugging Face also offers community GPU grants for side projects.
Scroll sideways to see all columns. With a keyboard, focus this table and use the arrow keys.
| Volume | Public repositories | Private repositories |
|---|---|---|
Base | $12/TB/mo | $18/TB/mo |
50TB+ | $10/TB/mo | $16/TB/mo |
200TB+ | $9/TB/mo | $14/TB/mo |
500TB+ | $8/TB/mo | $12/TB/mo; contact sales for custom pricing above 500TB |
Buying take
If you publish public models and demos, start on Free with CPU Basic hardware and pay only for the upgraded Spaces hardware hours you use; move to PRO when you want to host your own ZeroGPU Spaces.