Tinfoil
Private AI cloud that runs models inside secure hardware enclaves, offered as a chat app, an inference API and containers.
Tinfoil is a private AI cloud that runs open-source models inside secure hardware enclaves, so user data stays encrypted and the privacy claims can be verified cryptographically. Tinfoil is used through Private Chat on web and iOS, an OpenAI-compatible inference API for developers, or Tinfoil Containers for running Docker images in an enclave.
Best for Developers and organisations handling sensitive data who need cloud AI inference with privacy guarantees they can verify.
- Platforms
- Web, iOS, API
- Pricing model
- Subscription
- API
- Yes
- Made by
- Tinfoil Inc.
- Added
Last verified: . Information taken from tinfoil.sh.
What is Tinfoil?
Tinfoil is a security research lab that operates an AI cloud for private, verifiable computation. Its models run inside secure hardware enclaves, so data is encrypted directly to the enclave and, according to the company, cannot be read or shared by anyone, Tinfoil included; these guarantees can be checked cryptographically. Three products sit on this base. Private Chat is an assistant for the browser and iOS with projects, web search, document upload, speech to text and syncing between devices. The Private Inference API is OpenAI-compatible, open source and billed by token usage per model. Tinfoil Containers runs any Docker image inside an enclave for custom applications and AI workloads, billed hourly for CPU, memory and GPUs. The model catalogue consists of open-source models on GPU-backed enclaves, and enterprise plans are quoted individually.
Key Features
Enclave-Based Inference
Runs models on GPUs inside secure hardware enclaves so prompts and outputs stay encrypted to that environment.
Verifiability
Lets users cryptographically verify the privacy guarantees and check them against open-source code.
Private Chat
Provides an assistant with projects, web search, document upload and speech to text on web and iOS.
OpenAI-Compatible API
Exposes open-source models through an API that follows the OpenAI format and charges per token.
Tinfoil Containers
Deploys any Docker image into an enclave for custom or non-AI workloads that need the same protection.
Usage Controls
Applies per-model rate limits by usage tier and soft spending limits that can be set per API key.
Screenshots
Use Cases
- Confidential Chat
- Discuss sensitive personal or work matters with an AI assistant.
- Privacy-Preserving Apps
- Build an AI product whose user data the developer cannot read.
- Regulated Industries
- Give teams in defence, legal or similar sectors access to model inference they can audit.
- Protecting Intellectual Property
- Work on proprietary code or documents without exposing them to a model provider.
- Custom Secure Workloads
- Run your own containerised service on enclave hardware.
Things to know
- Spending limits are soft
- Requests already in flight when a limit is reached still complete and are billed, so the company suggests bounding concurrency if a hard cap is needed.
- Rate limits scale with spend
- API tiers rise automatically as lifetime paid usage grows; promotional credits do not count.
- Chat has hourly allowances
- Chat usage is capped by input and output token quotas counted per hour, which renew when the next clock hour begins.
- GPU commitment
- GPUs for Containers are priced by the hour but require a monthly commitment.
FAQs
What is Tinfoil?
Tinfoil is a private AI cloud from a security research lab that runs AI models inside secure hardware enclaves and offers a chat app, an inference API and container hosting.
How does Tinfoil keep data private?
Tinfoil encrypts data directly to a secure enclave where the model runs, and says the data cannot be accessed even by Tinfoil. The guarantees are cryptographically verifiable.
Does Tinfoil have an API?
Yes. Tinfoil's Private Inference API is OpenAI-compatible and open source, with pay-as-you-go pricing based on input and output tokens for each model.
Which models does Tinfoil offer?
Tinfoil offers a range of open-source models listed in its model catalogue, all running in GPU-powered secure enclaves.
How is Tinfoil billed?
Tinfoil Chat is a flat monthly subscription, the API is pay-as-you-go per token, and Containers are billed hourly for CPU, memory and GPUs, with GPUs requiring a monthly commitment. See the official pricing page for rates.
Does Tinfoil work on mobile?
Yes. Tinfoil's Private Chat is available on iOS as well as in the browser, with conversations synced across devices.
Can I cancel Tinfoil at any time?
Yes. Tinfoil says subscriptions have no long-term contract or cancellation fee, and service continues to the end of the billing period.
Who it's for
- Developers
- Enterprises
- Regulated industries
- Researchers
- Privacy-conscious users
Comments
No comments yet. If you have used Tinfoil, tell other readers how it went.
Alternatives to Tinfoil
| Tool | Free plan | Starting price | API |
|---|---|---|---|
 DeepSeekAI research company offering a chat assistant, a desktop harness and a token-billed API for its language models. |
— | — | Yes |
 APIParkUnified gateway for calling, authenticating, billing and monitoring AI models, APIs and MCP services. |
— | — | Yes |
ChatGPTAI chat assistant that answers questions, writes, creates images, runs deep research and helps with code. |
Yes | — | — |
 Duck.aiDuckDuckGo's free chat front end that anonymizes conversations with third-party AI models. |
Yes | — | — |
 Google GeminiGoogle's AI assistant for chat, research, and image, video and music generation, tied into Google apps. |
Yes | — | Yes |
Listed in the same categories as Tinfoil. A starting price appears only where we checked it on the maker’s pricing page; a dash means not checked or not published. All Developer API tools