NEAR AI announced on 15 October 2026 that Brave Leo, the AI assistant built into the Brave browser, now runs on private inference from NEAR AI Cloud. Users who select GLM 5.3 Flash get inference inside hardware-enforced trusted execution environments (TEEs) on Intel TDX and NVIDIA Confidential Compute.
The launch was also posted on X by NEAR Protocol, NEAR AI and Brave — credit via @NEARProtocol and @near_ai on X (posts verified through the X API).
How the privacy path works
According to NEAR AI’s blog:
- Leo encrypts each prompt in the browser with Oblivious HTTP, to a key that exists only inside the enclave. Brave’s relay strips the user’s IP before forwarding, so NEAR AI never learns who is asking and Brave never sees what they ask.
- Prompts decrypt only inside the enclave; memory stays encrypted during execution.
- The host OS, GPU operator and NEAR AI are locked out at the hardware level from reading the conversation.
- Every response is signed inside the enclave; hardware attestation proves the computation ran in a genuine TEE on NEAR AI’s open-source code, serving the chosen model.
Verification does not rest on NEAR alone. Before Leo’s browser receives the enclave’s key, Brave’s own open-source verifier checks the TDX quote against Intel’s signature and the GPUs against NVIDIA’s attestation service, and rebuilds NEAR AI’s code from public source to confirm it matches what runs. NEAR AI’s phrase: “No pass, no key.” NEAR AI Cloud also returns an Intel-signed token from Intel Trust Authority alongside raw attestation evidence.
GLM 5.3 Flash is described as an open-weights model under the MIT licence, so attestation can be tied to a model anyone can download and inspect.
From Nightly test to production
NEAR AI says Brave began deploying NEAR AI TEEs in November 2025, running DeepSeek V3.1 in Brave Nightly for early experimentation. This release extends private inference from the testing channel to the full production browser for every Leo user who selects GLM 5.3 Flash.
NEAR AI also claims Brave serves “over 128 million monthly active users.” That MAU figure is NEAR AI’s claim, not an independently audited Brave filing cited here. Other production workloads NEAR AI lists on the same cloud include Venice AI, Abound and the Government of Bermuda.
Quotes
Brendan Eich, Brave’s co-founder and CEO, said AI has raised the stakes while too much of it still runs on “‘trust me bro’ promises,” and that the NEAR AI integration brings “trust but verify” principles to AI chats in browsers: conversations sealed in the browser, opened only inside verified TEEs hosting a choice of open-source models, with privacy-proving code open to check.
Illia Polosukhin, co-founder of NEAR, called Brave extending confidential inference from NEAR AI to all their users “an exciting milestone,” stressing privacy and verifiability for a primary computing interface that handles personal information.
Why it matters
Browser assistants see the pages users read and the questions they would not want logged. Most cloud assistants protect that material with policy; TEE + attestation replaces “trust the provider” with a checkable enclave and a verifier Brave controls. For crypto-adjacent privacy stacks, that is a concrete product surface — AI inference that can sit next to wallet and DeFi habits without shipping prompts in the clear to a host.
On Solana-era infrastructure cadence elsewhere in crypto, see TSN’s Solana Alpenglow mainnet timing — a different protocol story, useful only as sector timing colour. Related NEAR Foundation and Sunrise Intents items are covered in companion TSN posts from the same research handoff.
What’s still unknown
- How widely Leo users will pick GLM 5.3 Flash versus other models
- Independent third-party audits of the attestation path beyond Brave’s verifier design
- Whether Nightly DeepSeek V3.1 remains available alongside production GLM 5.3 Flash
What this means
- Brave Leo GLM 5.3 Flash inference now on NEAR AI Cloud TEEs (Intel TDX + NVIDIA Confidential Compute).
- Brave’s verifier: no pass, no key.
- 128m+ MAU is NEAR AI’s figure.
- Credit: via @NEARProtocol and @near_ai on X.
Sources
- NEAR AI, “NEAR AI Cloud Brings Private Inference to Brave Leo”, 15 October 2026
- NEAR Protocol on X, 15 October 2026
- NEAR AI on X, 15 October 2026
- Brave on X, 15 October 2026
Related stories
Follow @tsnmediahub on X for breaking AI news as it happens.
Follow on X
