Input and output tokens
Requests in the selected range
Before discounts, exemptions, and credits
Credit balance
Current balance for this account
Input and output tokens
Requests in the selected range
Before discounts, exemptions, and credits
Credit balance
Current balance for this account

A look at zero data retention across 12 AI providers and services: what the promise covers, where it stops, and why verifiable privacy requires more than a contract.
Read postQuickstarts, SDKs, and API reference.
Compare model and infrastructure pricing.
Access the latest AI models without exposing your data.
Access the latest AI models without exposing your data.
Inference API access to powerful models, all running privately.
Run any containerized workload in a secure enclave.
Prices in USD.
Prices in USD.
All containers in your organization. Create new ones or manage running instances.
A config groups instances and secrets for a repository. The first instance or config secret creates the group automatically; setting up a configuration does not create an instance.
Storage owned by your organization. Volumes live on one host, attach only to containers on that host, and survive container deletion.
Store sensitive values like API keys and tokens. Secrets are injected into your containers at runtime as environment variables.
Manage SSH keys for accessing your debug containers. Adding SSH keys allows you to deploy containers in debug mode and SSH into the container to get access to the Docker logs.
Add and verify custom domains for your containers.
Configure credentials for private container registries to pull images during instance launch.
Create a verified model-weight artifact from Hugging Face. This prepares the weights for a container config; the model is deployed from the All Instances tab.
Tinfoil Bunkers are persistent, private cloud computers for your agents.
Storage owned by your organization. Volumes live on one host, attach only to containers on that host, and survive container deletion.
Publish and manage your organization's signed config versions under your namespace.
Config path
Config YAML
Renew the endorsement automatically
Manage your inference API keys
Admin keys provide programmatic access to your personal account, including access to billing and inference API keys. You do not need admin keys unless you're building advanced services with Tinfoil. Please read the documentation before getting started.
Admin keys have elevated privileges and can perform high-privilege operations including creating new API access keys. Never store admin keys in client-side code, browser storage, or any user-accessible location.
Your usage tier and the rate limits that apply to each model.
Higher tiers unlock higher API rate limits. Your account moves up automatically when the total of your paid invoices and credit purchases reaches each threshold.
Need higher or custom limits? Contact us.
Contact usAPI usage is limited per model in requests per minute (RPM) and tokens per minute (TPM).
How many containers you can run and how much storage you can allocate.
Higher tiers allow more containers, bunkers, and storage. Organizations with GPU access are automatically Tier 3.
What counts against your limits right now.
How many bunkers you can run and the resources each one gets.
Higher tiers allow more containers, bunkers, and storage. Organizations with GPU access are automatically Tier 3.
What counts against your limits right now.
Every bunker gets the same compute. You choose its workspace size when you create it.
Input tokens
Includes cache reads
Output tokens
Generated tokens reported by the model
Model requests
Excludes request-based services
Estimated API cost
Includes token-based and per-request services
Estimated cache savings
Already reflected in API cost; not an additional credit
Requests
Recorded usage events in the selected range
For the amount due after account adjustments, see Billing. For container costs, switch to Containers on this page.
Uncached input tokens
Input tokens not served from cache
Cache read tokens
Input tokens served from cache
Cache read ratio
Cached input ÷ total input; not a request hit rate
Estimated savings
Versus uncached input at current prices, before account discounts and exemptions
Only models with discounted cached-input pricing are included. Cache-miss reasons and cache-write counts are not recorded. Savings may be unavailable for models without pricing data.
Number of containers running over time.
Average number of running bunkers in each interval.
Review your current-period spending and usage.
Estimated savings · last 7 days
For models with discounted cached input, before account discounts and exemptions. Included in reported API cost.
Input and output tokens · last 7 days
View the Usage page for a breakdown by model.
Higher tiers raise your per-model requests and tokens per minute.
Review your balance, usage limit, and credit history.
Set an optional cap on API usage for each billing period. Access pauses when recorded usage reaches the cap.
This billing period
Limit per billing period
Leave empty or enter $0 for no limit. Usage resets each billing period.
Notify at % of limit
Send an email when usage crosses this threshold.
Redeem promotions and review discounts or credits for your account.
Enter a promo code to receive a discount or usage credits.
Review your subscriptions, billing periods, and payment methods.
View your invoices, payments, and receipts.
Recorded actions performed via admin API keys and the dashboard.
Infrastructure, configuration, billing, access, and key changes. Routine reads are shown separately.
Register apps that sign in Tinfoil users and call the inference API on their behalf.
You lose access to its resources until someone invites you again.
Input and output tokens
Requests in the selected range
Before discounts, exemptions, and credits
Credit balance
Current balance for this account

A look at zero data retention across 12 AI providers and services: what the promise covers, where it stops, and why verifiable privacy requires more than a contract.
Read postQuickstarts, SDKs, and API reference.
Compare model and infrastructure pricing.
Access the latest AI models without exposing your data.
Access the latest AI models without exposing your data.
Inference API access to powerful models, all running privately.
Run any containerized workload in a secure enclave.
Prices in USD.
Prices in USD.
All containers in your organization. Create new ones or manage running instances.
A config groups instances and secrets for a repository. The first instance or config secret creates the group automatically; setting up a configuration does not create an instance.
Storage owned by your organization. Volumes live on one host, attach only to containers on that host, and survive container deletion.
Store sensitive values like API keys and tokens. Secrets are injected into your containers at runtime as environment variables.
Manage SSH keys for accessing your debug containers. Adding SSH keys allows you to deploy containers in debug mode and SSH into the container to get access to the Docker logs.
Add and verify custom domains for your containers.
Configure credentials for private container registries to pull images during instance launch.
Create a verified model-weight artifact from Hugging Face. This prepares the weights for a container config; the model is deployed from the All Instances tab.
Tinfoil Bunkers are persistent, private cloud computers for your agents.
Storage owned by your organization. Volumes live on one host, attach only to containers on that host, and survive container deletion.
Publish and manage your organization's signed config versions under your namespace.
Config path
Config YAML
Renew the endorsement automatically
Manage your inference API keys
Admin keys provide programmatic access to your personal account, including access to billing and inference API keys. You do not need admin keys unless you're building advanced services with Tinfoil. Please read the documentation before getting started.
Admin keys have elevated privileges and can perform high-privilege operations including creating new API access keys. Never store admin keys in client-side code, browser storage, or any user-accessible location.
Your usage tier and the rate limits that apply to each model.
Higher tiers unlock higher API rate limits. Your account moves up automatically when the total of your paid invoices and credit purchases reaches each threshold.
Need higher or custom limits? Contact us.
Contact usAPI usage is limited per model in requests per minute (RPM) and tokens per minute (TPM).
How many containers you can run and how much storage you can allocate.
Higher tiers allow more containers, bunkers, and storage. Organizations with GPU access are automatically Tier 3.
What counts against your limits right now.
How many bunkers you can run and the resources each one gets.
Higher tiers allow more containers, bunkers, and storage. Organizations with GPU access are automatically Tier 3.
What counts against your limits right now.
Every bunker gets the same compute. You choose its workspace size when you create it.
Input tokens
Includes cache reads
Output tokens
Generated tokens reported by the model
Model requests
Excludes request-based services
Estimated API cost
Includes token-based and per-request services
Estimated cache savings
Already reflected in API cost; not an additional credit
Requests
Recorded usage events in the selected range
For the amount due after account adjustments, see Billing. For container costs, switch to Containers on this page.
Uncached input tokens
Input tokens not served from cache
Cache read tokens
Input tokens served from cache
Cache read ratio
Cached input ÷ total input; not a request hit rate
Estimated savings
Versus uncached input at current prices, before account discounts and exemptions
Only models with discounted cached-input pricing are included. Cache-miss reasons and cache-write counts are not recorded. Savings may be unavailable for models without pricing data.
Number of containers running over time.
Average number of running bunkers in each interval.
Review your current-period spending and usage.
Estimated savings · last 7 days
For models with discounted cached input, before account discounts and exemptions. Included in reported API cost.
Input and output tokens · last 7 days
View the Usage page for a breakdown by model.
Higher tiers raise your per-model requests and tokens per minute.
Review your balance, usage limit, and credit history.
Set an optional cap on API usage for each billing period. Access pauses when recorded usage reaches the cap.
This billing period
Limit per billing period
Leave empty or enter $0 for no limit. Usage resets each billing period.
Notify at % of limit
Send an email when usage crosses this threshold.
Redeem promotions and review discounts or credits for your account.
Enter a promo code to receive a discount or usage credits.
Review your subscriptions, billing periods, and payment methods.
View your invoices, payments, and receipts.
Recorded actions performed via admin API keys and the dashboard.
Infrastructure, configuration, billing, access, and key changes. Routine reads are shown separately.
Register apps that sign in Tinfoil users and call the inference API on their behalf.
You lose access to its resources until someone invites you again.