Inference Protocol Open Dev Demo
Our first Inference Protocol PR merged today! https://t.co/lLyVusJ1tJ
This PR focused on our first iteration of the Compute Marketplace—half of the Inference protocol.
Quick Points
1. Models and their corresponding ZK verify contracts are stored on-chain in a model registry.
2. As a result, the marketplace is trustless in nature. The User does not have to trust the Compute Provider to run inference on the correct model with the right input.
3. Privacy is preserved for output data. Since the User supplies a public key with each job they post, the Compute Provider can encrypt the output before posting to the chain.
4. Users set price they are willing to pay per inference
Join the open dev! Feel free to open issues, engage in discussions, or even submit PRs. Our devs are happy to collaborate as we will continue to dev in public.
Note: All $CHAT tokens will convert 1-to-1 to the Inference token in the near future. The timetable has not been announced yet. Learn more on our site or in our Telegram.
🛠️ Open dev of Inference is now live!
— Built as an EVM protocol for ease-of-access
— First modules focus on contest logic and compute proofs; follow the commits in real time
— Feel free to contribute, audit, or just watch the architecture take shape
https://t.co/HEVNbfaWgI
$CHAT
🚀 Inference Cloud is live!
– Finds the cheapest GPUs across top providers
– Spin up clusters and deploy any model or Docker image in minutes
– Unified dashboard, single API, zero platform fees
Start building → https://t.co/84IKjqNICS
Next up: the $CHAT-powered Inference protocol.
🛠️ Inference Cloud Video Demo
The Inference Cloud is set to go live this Friday! We’re on the last round of polish and bug fixes before release.
The platform will save ML teams incredible amounts of time by:
- Simplifying the search for cheap and reliable GPUs
- Single dashboard to spin up, deploy, and manage clusters across all major providers
Turning hours of work into seconds.
$CHAT
📅 Mark your calendar — June 16
Inference Cloud drops:
– GPU clusters online in minutes
– Deploy any open-source model or your own Docker image
– Cheapest rates, zero platform fees
Sneak-peek UI below. Open protocol development starts next.
$CHAT
$CHAT → Inference token
Holding $CHAT? No action required.
When Inference mainnet goes live, every $CHAT swaps 1-for-1 into the Inference token—100 % of the genesis supply. No new insiders, no dilution.
Supply is hard-capped; emissions are tamer than comparable protocols and tied only to live contests and verified compute.
Every token spent on network compute is burned, tightening supply as adoption grows..
Holders today form the foundation of tomorrow’s open-AI economy.
$CHAT rebrand is on the way with a new logo, fresh design, updated website, and the long-awaited token utilities.
I’m in the closed beta group as well, and things are looking solid so far.
If everything goes well, there’s no ceiling for a new ATH.
NFA, just a personal opinion.
VectorChat/Toffee beta testing - another evening
Spent last night talking to several characters inside the Toffee app (closed beta, invite only for selected).
And honestly? It felt more like a real conversation than anything I’ve seen from AI so far.
What stood out:
– natural rhythm
– emotional depth
– storytelling that pulls you in
– empathy that feels real
Conversations are absorbing. Deeply.
It felt like I was speaking with someone who understands emotion, builds tension, and knows how to tell a story.
They have tone, personality, presence — and that something you instantly recognize but can’t quite explain.
A few bugs? Sure. But the team reads feedback and reacts fast — that alone deserves credit.
More tests today. Still a few characters left.
More insights coming tomorrow! 🙂