Lightchain AI @LightchainAI

Your AI got worse, and no one can say why. Black boxes won't verify which model answers you. Lightchain tracks every LLM in the network, lets you dispute any answer, and operators put their $LCAI deposits on the line. Open intelligence, built to scale.

Open models are only half open. The network serving them needs to be open too. Lightchain lets any $LCAI depositor with a GPU serve LLMs like gpt-oss and Qwen3, and the community decides which ones run. Try the open network for AI: lightchain.ai

Bridging to Lightchain just got a lot simpler. New on the LCAI Bridge ⚡ Buy & Bridge: send ETH, receive native LCAI on Lightchain. One transaction, one signature. No more buying the ERC-20 first and bridging it separately. Transfer history that actually works, with direct links to every delivery on Lightscan. Redesigned bridge card and wallet tray. Under the hood: a small, open-source, verified contract swaps on Uniswap and hands off to Hyperlane in the same call. It holds nothing between transactions. Try it - bridge.lightchain.ai etherscan.io/address/0x9196…

The right answer alone isn't enough. AI compute needs an open standard where anyone can see the work behind every model's response. Lightchain keeps an onchain record of the LLMs, their jobs, and the $LCAI deposits that back the outcomes. Verify it yourself → workers.lightchain.ai

Lightchain takes the spotlight ⚡️ We're building open intelligence for the AI era. Compute you can verify, models you can trust, and a simple experience on your desktop machine.

Lunar Strategy @LunarStrategy

4 days ago

Project spotlight: @LightchainAI Lightchain lets people run AI models on their own machines and earn fees for answering requests. Its Desktop app brings that network into one interface: → chat in private, peer-to-peer rooms → manage a wallet → pay for AI answers with LCAI

3 15 45 4K 3

The open network for AI builders is expanding. Lightchain has a new growth partner: @LunarStrategy. We're excited to grow the developer ecosystem and the global market for verifiable compute. Onwards ⚡

One install instead of a tab for every corner of Lightchain. Rooms, wallet, paid AI inference and worker hosting now sit in one desktop app, with no account to create. Windows + macOS beta are live → lightchain.ai/download

🚀 Lightchain Desktop is LIVE in Beta Now available for Windows + macOS, bringing the Lightchain ecosystem into one desktop application: 💬 Encrypted P2P rooms — no central server 💳 LCAI wallet + send/receive 🔄 Ethereum ↔ Lightchain bridge 🤖 AI inference paid in LCAI ⚡ Run an AI worker from your machine 🔷 Create & monitor validators 📥 Download: lightchain.ai/download And there’s more coming 🛠️ Developer API + SDK → Testnet planned next week 📚 New developer docs → Planned alongside the testnet release 💬 chat-v2 → chat.lightchain.ai → Migration complete 💻 Lightchain IDE → In development & testing The Desktop release is currently beta. Builds are not yet code-signed and updates are manual for now. Try it, test it, and send us your feedback. 🔷

🏛️ DUNA Members — Governance is LIVE LCAI holders can now vote DUNA team members in or out of their positions. Once a proposal passes, anyone can execute it and update the on-chain member registry. The DUNA team handles the corresponding off-chain side. 🗳️ Threshold: 140K LCAI 📊 Quorum: 3% · 298.94M LCAI ⏳ Voting delay: ~1 hour 🗳️ Voting period: ~3 days 👥 Current members: 14 🔗 dao.lightchain.ai/duna/create DUNA governance. Community decisions. On-chain execution. 🔷

🔷 chat-v2 is LIVE on LightChain Mainnet — Public Beta 👉 chat-v2.lightchain.ai The first chat where no company decides who answers you. Workers claim requests directly from the chain, and every response comes with cryptographic proof of who produced it. ✅ Workers self-select on-chain — no dispatcher ✅ Proof on every answer — verify the worker, job, and committed answer ✅ Price before you send — real on-chain fee, starting at 0.02 LCAI/message ✅ Verifiable transcripts — anyone can re-check responses against the chain ✅ Private memory — stored locally in your browser ✅ End-to-end encrypted — the network only sees ciphertext 💰 Pay per message from a prepaid LCAI balance. Unused balance can be withdrawn anytime. The existing chat.lightchain.ai remains unchanged and continues working as before. 🚀 Try chat-v2 chat-v2.lightchain.ai

🚨 LightChain Mainnet Upgrade Reminder Tuesday, Sept. 22 at 13:59:26 UTC ⏱️ 2-second slots begin at slot 2,017,074. This is the second date of the upgrade already made — no new config or restart required. Before the cut, verify: 1️⃣ Validator client: ghcr.io/lightchain-pro… Digest: sha256:2b06848bfed0284a5fe67464d3d0faa1d651d00bb5f9b425a2ca046df4d6ed96 ⚠️ v7.1.2-lc.1 uses the old 6-second timing and will attest one slot late after the cut. Its attestations and sync-committee messages will be dropped. Only the validator container needs updating. Beacon nodes can remain unchanged. The swap takes only a few seconds. 2️⃣ Confirm your beacon node reports the fork: curl -s localhost:3500/eth/v1/config/spec | jq '.data.SLOT_TIME_FORK_ONE_SLOT, .data.SLOT_TIME_FORK_ONE_MS' Expected: "2017074" "2000" 🛑 Do not restart your node between now and one hour after the cut unless one of these checks fails. If a check fails — or you are not attesting today — reply here before 12:00 UTC Tuesday.

After the switch, only workers on the new image running the external profile will receive jobs. Setup: docs.lightchain.ai/docs/worker-an… How it works: docs.lightchain.ai/docs/worker-an…

Mainnet update for worker operators. Job assignment is moving from the dispatcher service to the chain; workers will claim sessions on-chain themselves. Existing setups keep working until the switch, which will be announced 24h in advance. No action required for chat users or LCAI holders.

Available now: - New image mainnet/worker:latest (build f24e37e): on-chain session claiming, preflight check, web search, all 11 whitelisted models - New gateway: …rker-gateway-v2.mainnet.lightchain.ai Pull the image, run preflight, and keep at least 10 LCAI on the worker address for gas.

DON’T CHASE THE NOISE. BUILD THROUGH THE QUIET. THE PEOPLE WHO STAY FOCUSED TODAY MAY BE THE ONES CELEBRATING TOMORROW. 🚀

🗳️ LCAI GOVERNANCE VOTING IS NOW OPEN The community has a voice. Now is the time to use it. Take a few minutes to review the proposal, understand what’s being presented, and cast your vote. Every vote matters when we're building a community-governed ecosystem. Read. Decide. Vote. 🚀 $LCAI #LCAI #LightchainAI #DAO #Governance #Web3

📢 New proposal submitted 🖊️ Signer Allocation for Multisignature Governance of Clearing House Wallet 📰 This proposal is a follow-up to the previously approved Lightchain AI DUNA — Clearing House Wallet proposal and establishes the initial authorized signer group for the Lightchain AI DUNA Clearing House Wallet. 💰 Collect your LCAI on the mainnet and get ready for voting in 24h from now! 🔗 Link to Proposal: dao.lightchain.ai/proposal/0xd21…

🚀 Excited to be part of the team helping manage the official Lightchain AI community on X. There’s a lot happening with LCAI, and we’re just getting started. I’ll be here to help keep the community informed, share important updates, and highlight the progress being made. Stay tuned. The journey continues. ⚡️ #LightchainAI #LCAI #AI #Web3

New AI models are whitelisted on Lightchain mainnet: llama3-8b, llama3-70b, gemma4:e2b, glm-4.7-flash, qwen3-coder-next, gpt-oss:20b, gpt-oss:120b, qwen3-vl:8b, qwen3-vl:30b, qwen3-embedding:0.6b, tts-piper Workers can now start running them. Our team is spinning up workers too, so you'll be able to use them on mainnet soon. workers.lightchain.ai/run-node

Testnet update. Prague/Electra: - Blob target/max 6/9, basefee update fraction raised. - BLS12-381 precompiles, EIP-7702 set-code delegation. Inference: - Answers stream from the worker as frames. - Vision on qwen3-vl:8b. - Speech settles audio as the response blob, same path as text. Chat: - Per-model live worker counts in the picker. - Liveness derives from SessionClaimed/JobAcknowledged logs, not registry eligibility, so dead registrations drop out. Costs: - Fee, status and full job lifecycle on chain. - Past deadline, claimTimeout refunds escrow, slashes the worker. Infra: - Two (2) GPU workers on an A100 80GB, staked independently. - Chain writes serialize per sender; geth reserves a sender to one txpool subpool. Roadmap: lightchain.ai/roadmap