Pinned Tweet
Introducing AQ: a multiplayer harness of harnesses.
• Run it on any cloud you already have, or use ours.
• Connect any coding harness you like: Claude Code, Codex, OpenCode, Cursor, Kimi, Grok, Devin and more.
• Work with your teammates live, way beyond sharing a session.
• 17 open-weights models, one click.
• Use it from anywhere: Mac, Linux, Windows, iPhone, Android.
AQ.dev is the easiest way to try new models on shared context/memory etc.
I'm sitting at a coffee shop outside the bay area/California, and next to me there's a student who's actually using Grok Bot.
I believe Elon that it's growing fast.
OK. What is happening? AQ.dev is blowing up 🎉
wow .. we should totally add it to @aqdotdev
We are releasing Ultramarine 1 Flash, a frontier-class model built for knowledge work. It is open-source and achieves similar performance to Fable 5, while being 60x cheaper.
It was post-trained by a team of 5 for ~$300k all-in. Available on API now, with weights releasing in 2 weeks.
Woah! Does this really beat Fable? You can try it on @aqdotdev
SubQ Snap powers contextual model routing on AQ.dev .. case study coming soon 💪
Introducing SubQ Snap, the first long-context decision model for real enterprise work.
Snap is the only decision model or LLM where compute per token stays constant as context grows.
• 384K tokens of multimodal context, 12x larger than Jev.
• 61.7 on Decision Index v0.2.1 vs Jev’s 57.9, outperforming on 13 of 16 enterprise benchmarks.
•Processes 100K tokens and returns a decision in under one second.
•$0.05/MM input tokens. Output is free.
Available today to anyone.
The first 10,000 sign-ups get 100M free input tokens.
We’ll likely hit capacity fast and will add users as we add compute. Compute offers welcome 🙂.
Latency will be variable as we tune the platform to this new model and category of users.
Model speeds are 80ms for minimal context, 250ms for 32k, 660ms for 100k, and 2.5 seconds for 384k.
Visit: subq.ai/
Love it. And if you want the same thing but on other models/harnesses, connect Dots to AQ and you can delegate to Claude, Grok, Muse, GLM, Kimi, etc. AQ.dev
Just a few weeks after Jev launches, every frontier model compu is competing. Competition has never been more fierce.
Introducing Microsoft-Decision-1, our new model for fast decision-making.
It delivers top performance on structured decision tasks, outperforming both LLMs and other decision models in latency and quality.
We’re already testing it across Microsoft for everything from incident response and quality control to scientific discovery.
Fun. If you use AQ.dev, you can do this with any model including open weight ones.
Finally. I’m telling you. No one will be coding locally. Experience it on AQ.dev
I'm officially sold on moving everything from running locally with Cursor, Claude, Codex, etc. to running the same models/effort levels, but with cloud computers.
Now moving my eval suite buildout for Frontier v5 in Claude Code to a Claude Cloud session.
Just asked Claude to give me instructions to put in a new session and it's jamming, next up, everything else.
Cursor is what first got me hooked, and now that Claude has this, it's hard to imagine working any other way.
Computer, phone, doesn't matter, they're all the same, everything is just a thin client to the cloud.
