Which models run in a TEE?

Private mode runs six open-weight models inside a trusted execution environment (TEE) that your browser verifies before each chat: GPT-OSS 120B, DeepSeek V4.1 Flash, Kimi K3, GLM-5.3, Gemma 4 31B, and Llama 3.3 70B.

The enclave runs on cloud GPUs in confidential computing mode. Its remote attestation proves which code runs, and your browser verifies it before it sends a word.

Private mode is the default of the app, and the free trial opens it with no account.

ModelMakerModePrice
GPT-OSS 120BChatOpenAIPrivate$0.09 in / $0.48 out per 1M tokens
DeepSeek V4.1 FlashChatDeepSeekPrivate$0.48 in / $1.08 out per 1M tokens
DeepSeek V4.1 FlashChatDeepSeekZDR$0.04 in / $0.10 out per 1M tokens
Kimi K3ChatMoonshot AIPrivate$3.00 in / $15.00 out per 1M tokens
Kimi K3ChatMoonshot AIZDR$1.00 in / $3.75 out per 1M tokens
GLM-5.3ChatZ.aiPrivate$1.35 in / $4.31 out per 1M tokens
GLM-5.3ChatZ.aiZDR$0.09875 in / $1.75 out per 1M tokens
Gemma 4 31BChatGooglePrivate$0.30 in / $0.75 out per 1M tokens
Gemma 4 31BChatGoogleZDR$0.125 in / $0.425 out per 1M tokens
Llama 3.3 70BChatMetaPrivate$1.31 in / $2.06 out per 1M tokens

More questions.

Every question

Get started. Think in private.

Start a private chat