QVAC

279 posts

QVAC banner
QVAC

QVAC

@qvac

Infinite intelligence. Local. Any Hardware. Peer-to-Peer Hyper Swarm. No cloud. No compromise. QVAC is the decentralized AI platform for humans and machines.

Katılım Nisan 2025
2 Takip Edilen10.7K Takipçiler
QVAC
QVAC@qvac·
Pears winner: Curva, by @bappeettt A P2P watch-party for football. Friends on different continents watch in sync, with chat, on-device AI translation and commentary, and gasless tipping. No servers. Built on @Pears_p2p, @WDK_tether, and QVAC. x.com/bappeettt/stat… github.com/louissarvin/Cu…
Capinho@bappeettt

Introducing Curva — a peer-to-peer World Cup watch-party. Two friends, two continents, one match. No streaming platform. No chat server. No custody. No cloud AI. Built on: Pears — 13 Holepunch primitives WDK — gasless USDT tips on Sepolia QVAC — on-device Bergamot translation

English
0
1
4
621
QVAC
QVAC@qvac·
The Tether Dev Cup winners are in. 157 participants shipped real apps on the Tether open-source stack: @QVAC, @WDK_tether, and @Pears_p2p. Overall + WDK: Tarkam QVAC: Scout Pears: Curva See what each project is about:
QVAC tweet media
English
8
16
59
5K
QVAC retweetledi
Paolo Ardoino 🤖
Paolo Ardoino 🤖@paoloardoino·
Tether AI + Hermes
QVAC@qvac

Hermes is an AI Harness from @NousResearch that is truly yours: you can shape its persona, it remembers you across sessions, and it gets better the longer you use it. We ran it fully local on QVAC. The persona, the memory, the skills it learns, all of it stays on your machine.

CY
10
12
121
25.2K
QVAC
QVAC@qvac·
Nice to see people building their own tools on QVAC. Here, a fully local AI for your Obsidian vault: chat with your notes, cited answers, no cloud, no data leaving your device. This is exactly what the SDK is for.
Samoth@samoth_ai

I love using Obsidian @obsdmd , but I always wanted to actually talk to my vault, ask questions across my own notes and docs. I just didn't want to hand any of that to a cloud AI, I wanted it fully local. So I built this. It runs on a local model with @qvac , nothing ever leaves your machine. Ask a question, get an answer grounded in your own notes with clickable citations. Semantic search, related notes, and you can even fine-tune a small model on your own writing so it answers in your style.

English
5
28
56
5.8K
QVAC
QVAC@qvac·
Built by the Tether Evo research team, it is a redesign of a large speech model that reads neural signals instead of audio. On real recordings it decoded full sentences word for word, and placed 4th of 466 teams in the Brain-to-Text '25 challenge. All in under 2 GB, on-device. It is early: a proof of concept, now in the QVAC SDK for developers to build on. A step toward giving a voice back to people who have lost theirs, with their thoughts never leaving their hands. How it works, and where it goes next: qvac.tether.io/blog/nothing-i…
English
4
2
15
973
QVAC
QVAC@qvac·
An AI that reads the words you are trying to say, straight from your brain, and writes them as text. No voice, no movement needed. For people who have lost the ability to speak, that changes everything. And nothing is more private than a thought. A system that decodes signals from your brain has no business running in someone else's cloud. It has to run on your own machine. That is exactly why BrainWhisperer runs fully on-device with QVAC.
QVAC tweet media
English
7
25
81
31.7K
QVAC
QVAC@qvac·
Sovereignty over your AI is this: you fully own it, you can reach it anywhere and any time, it costs you nothing to run, and no one can cut the link between your agent and its model. That only happens when the model is local. The brain is a file on your disk, not a service someone else can throttle, change, or take away. Want your own? The setup is a copy-paste walkthrough: start the model with QVAC, point Hermes at it, and start talking. Works on macOS, Linux, and Windows. Full setup guide: qvac.tether.io/blog/how-to-ru…
English
0
1
11
1.7K
QVAC
QVAC@qvac·
Hermes is an AI Harness from @NousResearch that is truly yours: you can shape its persona, it remembers you across sessions, and it gets better the longer you use it. We ran it fully local on QVAC. The persona, the memory, the skills it learns, all of it stays on your machine.
English
11
19
119
57.4K
QVAC
QVAC@qvac·
The Tether Developer Cup final four is now revealed. 157 participants. Four teams. Three countries. One cup. Congratulations to our four finalists.
QVAC tweet media
QVAC@qvac

The Tether Developers Cup is live: a global hackathon that runs like a football knockout tournament. Teams of 1 to 4 build a football-themed project on the Tether open-source stack, pick a nation to represent, and survive five rounds until one team lifts the cup. Three tracks: - @qvac for on-device AI - @Pears_p2p for peer-to-peer apps - @WDK_tether for self-custodial wallets 8,000 USDt in prizes. Free, worldwide. Register June 28 to July 6, live Final July 15. Register with the link below ⬇️

English
3
6
20
6.1K
QVAC
QVAC@qvac·
Your laptop's AI forgets long conversations because it runs out of working memory, not intelligence. QVAC team shipped an open-source fix: TurboQuant, a 5x memory compression algorithm for local AI, available in QVAC SDK. Read how it works: wired.com/sponsored/stor…
English
5
5
48
2.5K
QVAC
QVAC@qvac·
@adidshaft @Alibaba_Qwen LLM Judge for creative tasks and Pass check for reasoning tasks. We'll publish a more detailed article on the methodology soon
English
1
0
2
86
adidshaft
adidshaft@adidshaft·
@qvac @Alibaba_Qwen same device + nothing leaves the machine is the right setup. i'd want the quality score unpacked though: human rubric, model judge, or task checks?
English
1
0
0
73
QVAC
QVAC@qvac·
We benchmarked three @Alibaba_Qwen models, a 4B, a 9B, and a 35B MoE, running 100% on-device through the QVAC SDK. Same device (MacBook Pro M5 Max, 36 GB), same prompts, nothing left the machine. We chose models that can run on widely available consumer machines. Here is what we found ⬇️
QVAC tweet media
English
6
11
84
22.2K
QVAC
QVAC@qvac·
The reasoning set. Five problems, one exact integer answer each, auto-graded against a brute-forced key. Each problem was sent on its own, greedy decoding: "Solve this math problem. Think step by step and show your working. Then, on the final line, write exactly "ANSWER: " and nothing else after it. 1. In how many ways can 8 people be divided into two teams of 4, where the two teams are indistinguishable (unlabeled)? 2. In how many ways can 7 distinct books be distributed to 3 different students so that each student receives at least one book? 3. What is the sum of the first 50 prime numbers? 4. Let F(1) = F(2) = 1 and F(n) = F(n-1) + F(n-2) for n >= 3. What is the remainder when F(100) is divided by 1000? 5. How many ordered pairs of positive integers (a, b) with 1 <= a <= 100 and 1 <= b <= 100 satisfy that (a + b) divides (a * b)?" Answers: 35, 1806, 5117, 75, 170.
English
1
0
3
467
QVAC
QVAC@qvac·
The animation task (a walking robot, HTML + CSS): "Create a single, self-contained HTML file (inline CSS only, no external assets, no images, no JavaScript required) that animates a small robot walking in place across a simple outdoor scene. Requirements, follow all of them: - The robot must have a head, a torso, two arms and two legs, built from HTML/CSS shapes. - The legs and arms move in a believable, smooth, seamlessly looping walk cycle using CSS @ keyframes (opposite arm and leg swing together, like real walking). - Add a simple background: a ground line, sky, and one or two details (a sun, hills, or clouds). - Tasteful, coherent colors. Clean composition. The whole thing should look intentional and polished, not a rough sketch. - The animation must loop forever with no visible jump. Take your time and make it genuinely good. Output ONLY the HTML document, starting with and ending with . No explanation before or after."
English
1
0
2
631

Keşfet

@bappeettt @Pears_p2p @WDK_tether @0zLin @NousResearch @Dudung_gg @curioswhispers @adidshaft