Private Intelligence
Our products
For your private thoughts, projects, and apps.
For your thoughts
Private Chat
Chat with powerful AI assistants while keeping your conversations private. Available in the browser and on iOS.
$20/month
For your projects
Private Inference API
Build AI applications that keep all data private. Our API is OpenAI-compatible, open-source, and fully verifiable.
Usage-based pricing
For your applications
Tinfoil Containers
Run any Docker image in a secure enclave. Bring privacy and verifiability to applications and custom AI workloads.
$20/month + usage
Industry highlights.
Tinfoil in case studies, research, and collaborations.
Featured in a Llama case study as the only multi-GPU infrastructure offering production-ready, verifiably private AI.
Collaborating with Red Hat on open-source confidential AI infrastructure for private inference.
Security audit and webinar collaboration on TEE vulnerabilities and confidential computing best practices.
Our story on building audit-ready, verifiably private AI with Ubuntu as the foundation for multi-platform confidential computing.
Industry Programs and Collaborations
Infrastructure built for privacy.
Tinfoil makes your AI workloads secure, verifiable, and private.
SOC 2 Compliant
Open Source
“Running our own custom Docker container on Tinfoil Containers is a major unlock. It lets us run our full end-to-end system in trusted hardware using the same simple Python SDK we already use to call Tinfoil's embedding and LLM models. Serverless enclaves have finally arrived!”
Darya Kaviani, UC Berkeley
The Open Anonymity Project
“When building The Open Anonymity Project at Stanford and UMich, we were using Azure's confidential containers (ACI) which is a nightmare to set up correctly, from TLS certificate binding, hardware measurements, reproducible image digests, etc. We can do the same thing on Tinfoil Containers in <20mins with the nice attestation SDK, clear docs, debug mode, almost zero update down time, and transparent architecture that everyone can audit.”
Erik Chi, The Open Anonymity Project
“We have fast deployment cycles for servers that we run on Tinfoil TEEs to guarantee customer privacy. Tinfoil Containers makes the TEE deployment friction almost nonexistent and lets us iterate quickly. It's an important step towards the future where most ML workloads are secured by running on verifiably-private TEEs.”
Rudolf Laine, Workshop Labs
Case StudyWorkshop LabsWorkshop Labs built Silo, the first production-ready multi-GPU private post-training and inference stack for frontier models, powered by Tinfoil Containers. All customer data is processed inside hardware-isolated secure enclaves with customer-managed encryption keys, with less than 10% performance overhead on production workloads.Private Post-Training and Inference for Frontier Models
Case StudyPour DemainPour Demain, a Brussels-based AI policy think tank, used Tinfoil Containers to build a verifiable clean room for auditing frontier models. Their work shows the feasibility of running gray-box interpretability evals on a 744B parameter model inside confidential enclaves: the auditor brings the eval, the lab brings its weights, and both get cryptographic proof of exactly what ran, without the weights ever leaving the enclave.Auditing a Frontier Model Without Seeing Its WeightsPrivacy of local.
Power of cloud.Tinfoil runs AI models inside secure hardware enclaves. This gives you the privacy of local AI with the power of cloud computing.
| Feature comparison| | Other providersClosed Source Models | Local AIOpen-source & Custom Models | 
Open-source & Custom Models |
| ---------------------------- | ----------------------------------- | ----------------------------------- | -------------------------------------------------------------------------------- |
| Provable Zero Data Retention | Not included in Other providers | Included in Local AI | Included in Tinfoil |
| Cloud Convenience | Included in Other providers | Not included in Local AI | Included in Tinfoil |
| Setup costs | LowLow in Other providers | HighHigh in Local AI | LowLow in Tinfoil |
| Complexity | LowLow in Other providers | HighHigh in Local AI | LowLow in Tinfoil |
| Scalability | GoodGood in Other providers | PoorPoor in Local AI | GoodGood in Tinfoil |
| Zero Trust | Not included in Other providers | Not included in Local AI | Included in Tinfoil |
| Private Observability | Not included in Other providers | Not included in Local AI | Included in Tinfoil |
Other providers
Local AI
Other providers
Closed Source Models
- Provable Zero Data Retention
- Cloud Convenience
- Zero Trust
- Private Observability
- Provable Zero Data Retention
- Cloud Convenience
- Zero Trust
- Private Observability
- Provable Zero Data Retention
- Cloud Convenience
- Zero Trust
- Private Observability
* LowSetup costs * LowComplexity * GoodScalability
Local AI
Open-source & Custom Models
* HighSetup costs * HighComplexity * PoorScalability
Open-source & Custom Models
* LowSetup costs * LowComplexity * GoodScalability
Frequently Asked Questions
Have a security or privacy question?
Detailed answers on what Tinfoil can and cannot see, how secure enclave attestation works, what would happened with legally-compelled disclosure, and more.
Security and Privacy FAQRead the Security and Privacy FAQ
How does Tinfoil protect my data?
What does end-to-end privacy mean?
How is Tinfoil different from typical AI security approaches?
Is there a performance overhead when using Tinfoil?
Is Tinfoil compatible with closed-source models like GPT-5 and Claude?
What analytics and observability does Tinfoil offer?
Can I deploy my own AI models?
What type of workloads are suitable for Tinfoil?
Can I use Tinfoil with coding agents and other third-party clients?
Can I integrate Tinfoil with my existing application?
Can I keep server-side data and state?
Will Tinfoil eventually consider using fully homomorphic encryption?
Our Mission
Tinfoil was founded to create the private garden for thought.
As AI becomes more personal, omnipresent, and powerful, protecting the privacy of our conversations, thoughts, memories, and intellectual property is what preserves human relevance. AI should be an exoskeleton that empowers humans rather than replaces them. We are building the technical privacy safeguards to make this possible, using encryption, secure hardware, and other means to make it impossible for third parties to access your brain: a prerequisite for human empowerment and flourishing.