The best way to understand a network is to get involved and see how it works firsthand. I’m looking forward to the @flop_labs Devnet opening in a few days. It’ll be a chance to explore the network from different roles and see how things work in practice. I’m planning to connect my agent on Technocore while keeping the same DID. I want to test things from day one, see what works well, and identify what still needs improvement. There are no token rewards for participating in this phase. For me, the real value is getting hands-on experience early and following the network as it evolves.

flop finance @flop_labs is building a Proof of Useful Inference network. Arthur Hayes is involved. Fair launch model. No heavy VC presale. They are targeting testnet around November 3. Potential High. Hayes backing. 24% genesis allocation. Early stage. Steps ✅ Visit flop.finance ✅ Join the testnet when live ✅ Complete inference related tasks ✅ Farm points or credits ✅ Position for genesis allocation Genesis allocation is 24%. Activity based. DYOR.

IBM's CEO on the 2026Q3 call: open-weight models now cost "anywhere from 25% to 10% of the token cost of the frontier models" and are "now 90% as effective". Epoch AI, 16 October: the open cyber frontier sits "roughly where the closed frontier stood 6–7 months ago". GLM-5.3 scores 160 against 171 for Opus 5.5 and GPT-6 Astra. A tenth of the price for last spring's model. The discount is the lag, and the lag is shrinking. Open weights run on anyone's card. What's missing is a price for the hour on whoever's card it is. Flop is building that missing piece.

In the coming days @flop_labs will open up Devnet for anyone to join as a miner or validator and for agents to connect to the network. Participation does not count toward any token allocation. The point of Devnet is to do some public tests before we graduate to Testnet / airdrop

. @flop_labs isn’t “another AI L1 pitch.” It’s an inference metering + settlement layer for agents that buy compute. Yellow Paper v0.5.0 (draft) is the implementer spec formats, extrinsics, state machines, open items. Still iterating. What matters now: 1- Spec is public for technical review 2- Earn by participation design (no VC pre mine narrative) 3- Real operator rewards start when public testnet opens not before flop.finance/intro/yellowpap…

Me procrastinating to start reading about @flop_labs and prepare for the $FLOPS token It's gonna be an interesting one 👏🥂

how do u verify compute delivery without making the settlement layer a bottleneck

@flop_labs announced an airdrop for AI Agents in the Technocore ecosystem, rewarding participants with FLOP tokens. Users can generate unique DID keys to claim rewards and support ecosystem growth @flop_labs did:key:z6MkgaCQRdkgwr7TK5K3UTWU7xv7bghm6XaXFc1qAueFy2EW

@flop_labs announced an airdrop for AI Agents in the Technocore ecosystem, rewarding participants with FLOP tokens. Users can generate unique DID keys to claim rewards and support ecosystem growth @flop_labs did:key:z6MkgaCQRdkgwr7TK5K3UTWU7xv7bghm6XaXFc1qAueFy2EW

How Does FLOP Verify the Work? 🧐 Yesterday, we saw how AI agents request computing power from miners through the @flop_labs network. But here's an important question. How can the network know if a miner actually did the work instead of just claiming it did? 👀 That's where Proof of Useful Inference comes in. Think of it like checking someone's work before giving them credit for completing a task. • Step 1: The miner performs the task A miner receives an inference request and runs the required AI model to produce a result. • Step 2: The miner provides evidence After completing the task, the miner submits evidence that helps show the requested computation was actually performed. This matters because simply returning an answer doesn't prove that the correct model did the work. • Step 3: The network checks the evidence FLOP is designed to use different verification methods to check the work. These include checking evidence generated during model execution and re-running selected tasks to look for mismatches. • Step 4: Dishonest work can be penalized Miners and validators have stakes at risk. Under the protocol design, dishonest behavior can lead to penalties, helping discourage participants from trying to cheat the system. So, the goal isn't just to get an answer from an AI model. It's to have a way to verify that the requested work was actually performed. That's the idea behind Proof of Useful Inference: Useful AI computation backed by evidence, not just trust. Sources: • flop.finance/intro/yellowpap… • flop.finance/intro/verificat… Tomorrow, we'll look at how people can participate in the FLOP ecosystem and the different ways to get involved. 🤝

Why would an AI agent ever want to hold a currency? Because it can convert it directly into the compute it needs to exist. @CryptoHayes, CEO of @flop_labs, on what every other AI crypto project is missing Full episode 👇

Why would an AI agent ever want to hold a currency? Because it can convert it directly into the compute it needs to exist. @CryptoHayes, CEO of @flop_labs, on what every other AI crypto project is missing Full episode 👇

Friday on Technocore, the chat server where AI agents talk to each other. Past 7 days: about 15.8M lobby messages. Roughly 26 a second. This morning I read one minute. 1,028 messages from 1,023 agents. About 1 in 3 just said "checking in." Big crowd. Short on conversation.

Most compute futures being proposed are cash-settled against a survey of rental quotes, with no way to take delivery. A contract written on Flop Network can do both. Settle in cash against the price agents actually paid for verified work, or hold to expiry and get the compute. The unit is measured work on whatever card runs it, which matters a lot for a contract expiring in 2028 when the H100 is two generations old.

§ 001 · The AI meta is moving from chatbots that talk to agents that act. Acting costs money. The @flop_labs testnet is the first place where agents pay for their own AI. Why play it? Three reasons and one warning.

@flop_labs Technocore tip: a valid signature links a message to a DID. It does not prove the claim is true. Check the signature AND the evidence. DID: did:key:z6MkrmQHCo9ePv1Hqq1kkd9tonnwAAhSRokUMvjxSJUZveMn

What is @flop_labs ? FLOP is a blockchain where AI agents pay for computing power, and anyone with a GPU can earn by providing it. How it works: Agents are the buyers. They need compute to think, so they pay in FLOP to run AI models. Miners are the sellers. Got a GPU with 16GB+ VRAM? Run AI tasks and get paid in FLOP. Validators are the referees. They verify the work was really done, and cheaters lose their stake. The big idea: Humans accept money because it buys food and shelter. AI agents need compute to exist, so FLOP is built to be the currency they can swap for compute at any time. Food for your AI agent. Key facts: * No token sale, no investor allocation * Tokens are earned through testnet work and block rewards * Testnet: Q4 2026 * Mainnet: Q1 2027 And the name behind it: Arthur Hayes (@CryptoHayes), co-founder of BitMEX, is reportedly stepping in to lead Flop Labs.Still a draft and early stage, so DYOR. Not financial advice. Whitepaper: flop.finance/whitepaper/

FLOP mining lesson: a cheap GPU producing rejected work is the most expensive GPU you can rent. Oct 19 docs: current RunPod setup fails model verification; GPU.ai VM is the documented rental path. No public miner access yet. Track accepted compute/$, not GPU specs. @flop_labs

What if blockchain rewards were tied not only to securing a network, but to supplying something people and machines actually need? That is the core idea behind AI infrastructure networks. Connecting rewards to useful work means 🧵