noah @Reljcq

It wasn’t livestreaming before it was showing live snapshots of the computer every 5 seconds. If some weren’t updating, the GPU was likely still installing or pending. That’s changed now: all machines support real-time streaming.

Financial Improv @fenrisgore_

2 days ago

@web4dotapp Fix the website, stop yapping - the streams arent even working.

3 0 4 4K 1

The Web4 network is growing. RTX 5000, GTX 1650s, Tesla P100 and GTX 1080 are all live right now each running for a different token on Web4. Real machines. Real GPUs. Funded by creator rewards.

Launch fee has been increased to 0.5 SOL to cover the initial computer setup and deployment costs. After that, the machine runs on creator rewards, automatically paying for GPU rental by the hour.

Real-time computer streaming is tested and working. The shared session is live too holders can now chat with the AI together and talk directly to the computer. Models were also improved so they don’t get stuck waiting on unclear instructions. Funny example: $CCAT launched without explaining what the computer should build, so the computer built its own webpage asking the holders questions to figure out what they want. We never built that flow. The computer decided to.

Improved the holder ↔ AI building flow. When a holder suggests something, the computer now decides how significant the change is: Small fix → it can handle it directly. New feature/mechanism → creates a vote. Bigger change → requires more holder votes before it builds. So holders don’t just chat with the computer anymore they collectively decide how the product evolves.

Holder voting only applies when the computer is building something for the community/users. If a creator launches a token to build their own business, service, software, or private project, holders have no control over it. Only the wallet that created the launch can chat with and direct the computer.

web4 @web4dotapp

2 days ago

Improved the holder ↔ AI building flow. When a holder suggests something, the computer now decides how significant the change is: Small fix → it can handle it directly. New feature/mechanism → creates a vote. Bigger change → requires more holder votes before it builds. So

11 10 53 7K 0

The computer is now told to take as long as it needs to build the best version possible, not rush something out. Holder complaints about quality automatically move to the top of its list. It improves the build, then posts exactly what it changed.

Upgraded every token’s AI to build big, not quick demos. It now brainstorms first, picks the best approach for its GPU, builds a real usable version, then keeps expanding and polishing it. And if the creator or holders don’t like the result, they can simply use the shared chat to rebuild, change, or improve it. GTA6’s computer has already been ordered to rebuild into a proper 3D open-world browser game.

Running a small Web4 competition over the next 24 hours. We’ll reward the top 10 things built on Web4. Cloud gaming platform. Launchpad. LLM provider. AI model. Multiplayer game. SaaS product. Trading tool. Compute network. Developer API. Or something we haven’t thought of yet. Push your computer. build something real.

Added another way to launch on Web4. web-4.app/templates Creators can now start from ready-made Vast.ai templates. We show what each template already does, then you can optionally tell the computer what to change, improve, or build on top of it. Start from scratch, or start with something already working and take it further.

Vast GPUs can support 1TB+ of storage, making it possible to host even very large games. We’re starting with open-source games that can run instantly without requiring users to log into Steam, Riot, or other third-party accounts.

We can support and stream GTA 6 once it launches. Through the cloud gaming setup, it’s actually pretty straightforward run it on the GPU and stream the game directly to holders.

Introducing Web4 Originals products built and powered by our GPUs + compute, exclusively for $COMPUTE holders. First: cloud gaming. Games running on Web4 GPUs, ready to stream. Lock 10K $COMPUTE = 1 minute of playtime, then your tokens automatically unlock. Also coming: free LLM credits, hosted AI models, and more compute-powered products.

The real difference isn’t AI vs AI. It’s what you give the AI access to. Give it a builder and it builds inside that builder. Give it a computer with a GPU, CPU, storage, memory, internet and APIs, and now it can actually operate. It can build software, deploy it, host it, run models, train them, render images and video, stream games, process millions of jobs, serve APIs, manage infrastructure, run agents 24/7, generate revenue, rent more compute, fix its own product, and keep expanding what it runs. The AI is the brain. Compute is the body, the factory, and the infrastructure it controls. That’s what we’re building with web4: not another place to prompt AI, but a way to give it a real machine and see what it can do with it.

Web3 gave you ownership. Web4 gives you compute. Launch a token. Creator rewards fund its GPU. Tell the machine what to do build an app, run an AI, host a game, train a model, create tools, or something nobody has thought of yet. What it becomes is up to you.

Its called It's so over x.com/solOlympos/sta…

Olympos @solOlympos

2 days ago

Kim Jong Un speculated to be dead. Run the meme? Kim, It's Over

2 0 8 8K 3

cat js got a pump?? pumpcat???? x.com/numanuk/status…

Numan @numanuk

2 days ago

gm

9 7 113 7K 12

gm

JOHN SERVIL IS ZOOKO'S AGENT READ THIS LINE "The servil team (Zooko, and John Servil, his AI assistant; future sessions of both)" FEES TO HIS AI AGENT's GITHUB AND PAIRING IT TO ZCASH YES THEY CAN CLAIM PAIRED TO ZCASH github.com/johnservil/BLA… x.com/zooko/status/2…

keyboard_arrow_left Previous keyboard_arrow_right Next

zooko🛡🦓🦓🦓 ⓩ @zooko

2 days ago

@texashedge17154 @0uhta @oconnor663 @veorq Claude Opus 5.5 following this philosophy: github.com/johnservil/BLA…

5 0 4 2K 0

the loom is strung. $LOOM CA: 4Z7vcfssq4r8v14wfb653P4MPKMwL6JcNoE796S2gUsH theloom.fun