Trusted AI

Private execution. Verifiable results.

Run agents, private LLM models, and GPU jobs inside hardware-backed TEEs. Keep secrets private, and prove what ran.

Start buildingTalk to sales
PhalaGoogleAWS

 

 

 

 

 

 

CPU machine

GPU machine

Confidential VM

Confidential AI cloud

Move existing Docker Compose workloads into CPU or GPU confidential machines. Keep the deploy path familiar; make the runtime verifiable.

Check detail page
hardware quote
runtime measurement
verifier report
terminal · attestation.json

 

{

 

 

 

 

}

 

Attestation

Every result can carry proof

Instead of asking users to trust a cloud claim, Phala emits runtime measurements that software can verify.

Read attestation docs

Trusted by 5,000+ users

Trusted by industry leaders and developers worldwide.

Nvidia logoOpenRouter logoZ.AI logoOPPO logoVenice AI logoIntel logoNear logoUniswap logoFlashbots logoCrossmint logo

Products

Full service for AI privacy: agent sandbox, LLM, and GPU.

 

proof path
cloud
Runtime sandbox

Agent sandbox

Run agent tools, app servers, and Docker services inside TEE-backed runtime sandboxes.

Deploy sandbox

H200

US · 24 vCPU

141GB VRAM

Intel TDX + NVIDIA CC

$2.56/GPU/hr

B300

US · 12 vCPU

288GB VRAM

Intel TDX + NVIDIA CC

$5.63/GPU/hr

proof path
gpu
Confidential GPU

GPU marketplace

Reserve H200 and B300 confidential GPU capacity for private AI training and inference.

Open GPU marketplace

Confidential models

Private LLM models with real model choice.

OpenAI-compatible LLM endpoints, private prompts, and verifiable runtime state.

Explore LLM models
MoonshotAI
Qwen
Z.ai
Xiaomi
Qwen
Qwen
MiniMax
Z.AI
MoonshotAI
Z.AI
Z.AI
Qwen
MoonshotAI
Qwen
Z.ai
Xiaomi
Qwen
Qwen
MiniMax
Z.AI
MoonshotAI
Z.AI
Z.AI
Qwen
encrypted

MoonshotAI: Kimi K2.6

262K context

$1.09/M input

Check detail
encrypted

Qwen: Qwen3 Coder Next

262K context

$0.18/M input

Check detail
encrypted

Z.ai: GLM 5.1

203K context

$1.21/M input

Check detail
encrypted

Xiaomi: MiMo-V2-Flash

262K context

$0.10/M input

Check detail
encrypted

Qwen: Qwen3.5-27B

262K context

$0.30/M input

Check detail
encrypted

Qwen: Qwen3.5 397B A17B

262K context

$0.55/M input

Check detail
encrypted

MiniMax: MiniMax M2.5

197K context

$0.20/M input

Check detail
encrypted

Z.AI: GLM 5

203K context

$1.20/M input

Check detail
encrypted

MoonshotAI: Kimi K2.5

262K context

$0.60/M input

Check detail
encrypted

Z.AI: GLM 4.7

131K context

$0.85/M input

Check detail
encrypted

Z.AI: GLM 4.7 Flash

203K context

$0.10/M input

Check detail
encrypted

Qwen: Qwen3 Embedding 8B

33K context

$0.01/M input

Check detail

All-in-one confidential compute platform for AI workloads.

Platform

Built for private AI work

Write code, dockerize, and deploy it as trustless TEE apps.

marvin@Mac ~/ai-agent % claude code

Claude Code

bun ‹ claude

 

 

 

 

 

 

 

Proven at Scale

Built for enterprise security and regulatory requirements.

Building with confidential AI

0+

Users

Runtime proofs generated and checked

0+

Daily Attestations

Near-native confidential GPU execution

0%

TEE Performance

Total VMs

Live network source from Dune

Open source

Confidential model tokens/day

2026-05-07

843M

Crawled from Phala's OpenRouter provider chart during server render.

daily tokens10 active models
View token source

Enterprise-Grade Compliance & Security

Deploy confidential AI with confidence. Phala is SOC 2 Type I certified and HIPAA compliant, with ISO 27001 certification in progress and privacy-by-design controls aligned with GDPR.

Visit Trust Center
SOC 2 Type I Certified
HIPAA Compliant
ISO 27001 In Progress
99.9% Uptime SLA
GDPR Compliant Processing
24/7 Enterprise Support

FAQ

Common Questions & Answers

Find out all the essential details about our platform and how it can serve your needs.

1

What is Trusted Execution Environment (TEE)?

TEE is a secure area inside a processor that protects code and data from the operating system, hypervisor, and other applications.

2

How does confidential AI protect sensitive data?

Sensitive data and AI models remain private during processing by running inside hardware-backed secure environments.

3

Is Phala compatible with existing AI frameworks?

Yes. Phala supports existing Docker services and popular AI frameworks including TensorFlow, PyTorch, and Hugging Face.

4

What are the performance implications?

Confidential GPU workloads typically target near-native performance, with roughly 5-10% overhead depending on workload and hardware.

5

How can I verify the security of my AI workloads?

Phala exposes cryptographic attestations so users and systems can verify the workload and runtime state.

6

How do I get started?

Install the Phala CLI, deploy a Docker workload, then inspect status, logs, and attestation from the command line.

Start building

Build AI you can prove.

Deploy private workloads, verify execution, and scale from models to GPU jobs.