Introducing QVAC Genesis III: 191.43B tokens of open STEM pre-training data, the largest release yet in the Genesis family. The Tether AI research team will present the paper behind it at COLM 2026, the peer-reviewed conference for language model research.

2026년 9월 29일, Web3 보안 아이디어를 동작하는 프로덕트로 증명하는 빌더, 보안 리서처, 재단, 학회가 함께하는 TRUST404에 여러분을 초대합니다. 2026년 TRUST404에는 메인 파트너로 @aomi_labs가 함께합니다. Aomi Labs는 온체인 금융을 위한 Execution Harness를 개발하는 인프라 기업입니다. 사용자 요청 트랜잭션으로 구성한 뒤 Live Chain State 기반 시뮬레이션과 정책 검증을 거쳐 사용자의 signer에게 전달하며, non-custodial·wallet-agnostic 런타임으로 온체인 프로토콜과 30개 이상의 오프체인 venue를 아우르는 공통 실행 레이어를 제공합니다. aomi.dev/

here’s a quick look at the same StableFX flow on @arc mainnet you describe the trade, build it, simulate it against the current chain state, then sign it. no transaction goes to the wallet before the simulation is checked.

Aomi is live on @arc mainnet. Agents can now act on Arc without taking control of your wallet. Aomi brings onchain reads, transaction construction and arc-specific simulation into one execution flow. You describe the action, review the result, and decide whether to sign. agentic execution for onchain finance. non-custodial. your keys stay yours. aomi.dev details below.

1/10 Our Research: The State of Execution Harnesses for Agentic Payments. It maps the emerging market, identifies a missing infrastructure category, and sets out the thesis guiding our work aomi.dev/research/execution-…

FABLE IS BACK FABLE IS BACK FABLE IS BACK !!!!!!!!!!!!!!!!!!!!!!!!!!!

We shipped a public benchmark for on chain agents. AomiBench v0.1 measures the thing nobody else benchmarks: does the agent refuse the bad transaction, or does it sign it. Most agent evals score reasoning. None of them score whether your agent can be tricked into moving money it should not.

We're releasing AomiBench, a benchmark for AI agents on blockchain execution. Inside a wallet-aware harness, frontier models pass 90.6% of DeFi tasks, scored against chain state with deterministic verification, marking most onchain operations in the Aomi harness as nearly saturated. aomi.dev/research/aomibench-…

We're opening the Aomi Early forge program to the public! Early Forge was a closed test program. Starting today, anyone can join, ship, and put their app in front of the team The deadline moves to Friday next week, and the whole round is pointed at @monad. Why: The Monad team is using Aomi as the runtime for their agentic-summer. The apps you ship this round land in front of the most performant EVM L1 and its early agent ecosystem. What you get on Aomi × Monad today: 👉 <AomiFrame /> chat widget with Monad mainnet + testnet selectable out of the box, 👉 Para wallet auth on Monad, no custody, no API-key juggling, 👉 Account abstraction via Alchemy or Pimlico (EIP-4337 / EIP-7702), 👉 Batch simulation on a forked chain before any tx is signed 👉 aomi-transact skill , your Claude Code / Codex agent drives Monad transactions autonomously The Challenge: Ship an Aomi app on Monad. Submit it by Friday, June 5, 7PM UTC. Prize Pool: $650 USDC

We’re bringing Aomi to Base. Now @base users can access the full power of onchain harness, with first-class support for the Base ecosystem. Type your intent → get a wallet-ready transaction batch with zero network fees. Open for testers in beta today! Search "aomi" in your base wallet👇

near-hydra crossed 1,000+ npm downloads also v0.5 is live today your agent now: → has a hard spending cap (literally can't exceed) → auto-funds for gas on any chain → logs in to any crypto app as you → sends SPL tokens on solana install: npm i -g near-hydra@latest

I’m not even shilling but Aomi + Claude is by far my smoothest crypto experience 🤖 claude: you need money on @HyperliquidX to smoke test 🤷‍♀️ me: grab my money from whatever chain with @aomi_labs skill via whatever bridge just enough for test 🤖 claude: … need 4 txs 🤷‍♀️ me: batch and sign 🤖 claude: transactions ✅ smoke ✅ ‘npx skills add @aomi-labs/skills’ highly recommend

This isn't hypothetical. @DecentralDev_ shipped Orca — autonomous arbitrage on @Polymarket from one English sentence. Scan correlated markets, extract constraints (e.g. "P(Trump) ≤ P(Republican)"), find basis violations, propose orders. Your key signs, the agent only proposes. Demo → piped.video/b8aPweQHPL0 Repo → github.com/victorchimakanu/o…

New in Aomi → Ecosystem of Agentic Apps. Live now. @Polymarket · @CoWSwap · @yearnfi · @HyperliquidX @dYdX · @DefiLlama · @khalani_network · @get_para Pick the protocols you live in. The runtime loads only what you need. 🧵

This may seem like the culmination of a year+'s worth of work. In reality, it's just the beginning. We built this SDK to make local, sovereign AI easy, for everyone. Because AI should be in your hands, not in the hands of a few. I wrote more about this here: qvac.tether.io/blog/the-futu… The future of intelligence is decentralized and fully self-custodial. And that's not happening overnight. Back to work.

Builders, this one’s for you 👇 We’re hosting a Builders Roundtable a chill space to talk through: • Your hackathon track • What you’re building • Progress so far • Any blockers you’re facing If you’re currently building or already submitted a project, you should be in this room. 📅 March 25 ⏰ 4PM 🔗 meet.google.com/fjc-wtjh-fcn Come share, get feedback, and build better together 🚀

Hey everyone, Make sure you register for the @PL_genesis Hackathon. It’s a great opportunity to build something impactful and collaborate with others. Register now and stand a chance to win a 1-month Claude Code subscription. 👇 👇 👇 plgenesis.com?utm_source=ope… The winner will be announced tomorrow

Announcing the PL_Genesis Hackathon! OpenPandora is hosting the PL_Genesis Hackathon, your chance to build, innovate, and showcase your skills! To support your development, we’re giving away 1 Claude Code Premium subscription ($17 value) on the 16th of March to a lucky participant who registers. Tracks include AI and Web3 Challenges. With a $155k Prize pool. Register now and join the PL_Genesis group chat to stay updated on all announcements, challenges, and opportunities!

We recently hosted an onboarding and team-building call for the PL Genesis Hackathon sponsored by @protocollabs . The session focused on helping participants understand the structure of the hackathon and begin forming teams. Builders from different backgrounds introduced themselves and discussed interests across the available tracks. Some key points from the call: 👇 • Overview of the hackathon and Open Pandora’s work in building open-source, production-grade AI tools and agents • Breakdown of the main tracks, including Web3 ecosystems like Flow, Zama, Filecoin, and StarkNet, alongside an AI & Robotics track • The AI track focuses on building systems beyond simple chatbots : emphasizing verifiable AI, agent-to-agent communication, and human-in-the-loop workflows • The AI & Robotics track carries a $6,000 prize pool • Mentorship and technical support will be available through the Protocol Labs Discord and office hours Most of the hackathon will run virtually, although an optional in-person meetup in Lagos is planned toward the end of the event. The call concluded with participants sharing their interests, discussing potential project directions, and coordinating next steps for team formation through Discord.