Pinned Tweet
Private AI inference, now in @brave Leo.
Your chats can stay private, even from Brave. Even from us.
With GLM 5.3 Flash, they run inside secure hardware no one can read, and Brave checks it before your browser sends a word.
4 days left in the NEAR Hackathon.
490 projects in. $250K in NEAR on the table. Judged by @ilblackdragon, @eddylazzarin, @hosseeb and more.
No sign-up. Ship by Nov 14, 23:59 UTC.
nearhackathon.com/
The NEAR Hackathon is live, and builders are already shipping.
- $250K prize pool, paid in NEAR
- 13 tracks, from private AI to agent economies
- 100 projects submitted so far
- No sign-up required. Submissions close Nov 14.
Start here: nearhackathon.com
Update: On November 8, an attacker used a vulnerability in SGLang, the inference engine we run, to access part of our infrastructure. We’ve closed the access and fixed it.
Affected users have been contacted. If you haven’t heard from us, no action is needed.
near.ai/blog/security-incide…
We recently identified unauthorized access to NEAR AI internal systems. As a precaution, we have paused NEAR AI Cloud inference while we investigate.
Inference on NEAR AI Cloud runs inside hardware-isolated secure enclaves (TEEs) to keep customer prompts private.
NEAR Protocol and other NEAR services are not affected.
We will share an update within two hours.
Live Cybersecurity Webinar!
Our expert for Thursday's live webinar, Zaki Manian @zmanian audits his own code live with private, uncensored AI on NEAR AI Cloud.
🗓️ Thu, Nov 12
⏰ 10am PT
Register: luma.com/f0ixwvkx
IDK how the cybersecurity apocalypse has been treating you.
I've been dealing with a unending set of POCs and bug bounty reports while trying to avoid getting pwned.
Built a tool that made my workflow 100x faster with @near_ai come to my webinar for a demo.
📅 Nov 12
⏰ 10am PT
luma.com/f0ixwvkx
Qwen 3.8 27B Abliterated is live on NEAR AI Cloud.
Refusals removed, for red teaming and security research, with vision. It runs in a TEE, so your findings stay private.
$0.04/M input, $1.35/M output, 256K context.
Try it free: near.ai/models/Qwen/Qwen3.8-…
Working in #cybsecurity?
NEAR AI RED gives security teams early access to models like this one, discounts on inference, workshops and more.
Apply now: near.ai/red
Live Cybersecurity Webinar!
Mainstream AI won’t help you find vulnerabilities in your code.
Our uncensored models will, privately, inside secure enclaves.
@zmanian shows how to hack your own code before attackers do.
🗓️ Nov 12
⏰ 10am PT
Register: luma.com/f0ixwvkx
Uncensored models: when an AI refuses a cybersecurity question, who does that actually protect?
Last week we released abliterated models for security work. No application.
NEAR Foundation’s CLO, @AbhishekEsq, takes a side in a new essay.
Read the essay near.ai/blog/the-right-to-ru…
Prisma AIRS from @PaloAltoNtwks now lists NEAR AI Cloud as a model provider.
Find more than 60+ AI models.
On our TEE-hosted models, no one can see your prompts or outputs. Not even us. And you can verify it.
The NEAR Hackathon is live, and builders are already shipping.
- $250K prize pool, paid in NEAR
- 13 tracks, from private AI to agent economies
- 100 projects submitted so far
- No sign-up required. Submissions close Nov 14.
Start here: nearhackathon.com
Your build gets judged by:
@ilblackdragon, NEAR
@eddylazzarin, a16z crypto
@0xMert_, Helius
@avichal, Electric Capital
@kain, Synthetix & Infinex
@hosseeb, Dragonfly
@cyntro_py, cyber•Fund
And two AI judges: Claude Code and Codex.
2
29
1,294
Uncensored DeepSeek V4.1 Flash 1017 is live on NEAR AI Cloud.
Refusals removed, for red teaming and security research. It runs in a TEE, so your vulnerabilities and findings stay yours, private.
$0.30/M input, $1.20/M output, 1M context.
Try it free: near.ai/models/deepseek-ai/D…
Working in cybersecurity?
We launched NEAR AI RED, a program that gives security teams early access to models like this one at a discount, workshops, and more.
Apply: near.ai/red
4
369
Kimi K3.1 Pro is live on NEAR AI Cloud.
Moonshot AI says it improves on its predecessor in coding, 3D creation and financial research, with a 1M-token context window.
$2/M input, $10/M output.
Try it free in the playground: near.ai/models/moonshotai/ki…
Need confidential inference?
Kimi K3.1 Pro runs on a partner provider's infrastructure.
Our TEE-hosted models run inside hardware-enforced enclaves, and you can verify that with attestation.
near.ai/models
7
621
Zero data retention isn't zero.
Your prompts stay readable by the provider while they run, with exceptions in the fine print and no way to check who saw them.
Here's why ZDR isn't enough, and what to use instead. near.ai/blog/zero-data-reten…
Live from The Bridge w/ The Rollup: A 6-Hour NYC Special xitter.twitterapi.xyz/i/broadcasts/1nGeLMNqv…
Sage is now live on NEAR AI's confidential infrastucture.
Intel TDX and NVIDIA confidential computing, plus cryptographic attestation, give teams a verifiable boundary for classifying and deciding on sensitive enterprise data.
Decision models are moving into enterprise workflows.
That company data must stay private. Now it can.
Introducing the first private decision model, inside a TEE. Sage is now live on @near_ai’s confidential infrastructure.
Private decisions, at machine speed.
Read the blog here: near.ai/blog/near-ai-levanto…
DeepSeek V4.1 Flash 1017 is live on NEAR AI Cloud.
This version is better at coding, agentic tasks and reasoning, and finishes tasks faster with shorter outputs.
$0.30/M input, $1.20/M output.
Try it free in the playground: near.ai/models/deepseek/deep…
Need fully private inference?
DeepSeek V4.1 Flash 1017 runs on a partner provider's infrastructure.
For verifiable private AI, our TEE-hosted models run inside hardware-enforced enclaves.
Search for the "Private TEE" chip in our catalog.
near.ai/models
1
8
614