AQ retweeted
this is big 💥
Hinoki is the only open-access LLM built for offensive security, and you can now run it on @aqdotdev.
dm me if you want to run @hinokisecurity this weekend, and we’ll get you off the AQ waitlist.
Launch pricing: 5% off through November 1.
Per 1M tokens:
Input $0.19
Cached input $0.152
Output $0.76
For code and systems you own or are authorized to test.
aq.dev/launches/hinoki/
Then it fixes what it finds. Every finding comes with severity, confidence, file and line, the evidence, and a patch you can review, test and merge in the same thread.
No scan proves code is secure. It is the sharpest first look you can give your code.
On CyberGym it scores 88.4%, ahead of OpenAI's gated GPT-5.5-Cyber at 85.6%. Post-trained from DeepSeek V4.1 Flash for offensive cyber performance.
General coding models are trained to build things. Hinoki is trained to find the ways they break.
What it does on offense:
Triage your CVEs: which of the forty advisories are actually reachable from your inputs, and how.
Write the exploit and run it against your own app, so you know which patch ships today.
Hunt logic flaws, auth gaps, injection chains and RCE paths that scanners never see.
AQ retweeted
I have wanted this one for a while.
@DevinAI now runs inside AQ. @cognition's terminal agent on an always-on machine, in an isolated worktree, in a session my whole team can open.
The thing I care about is not "another logo in the list." It is that harness choice stops being a team decision. One engineer runs Devin, another runs Codex, same repo, same workspace list, everybody sees everybody's session. You keep the diff that wins.
Your Devin subscription pays for Devin. We add nothing on top.
Launch page in the reply.
Try it: connect Devin CLI in AQ today. Free personal sandbox, nothing to install.
aq.dev/launches/devin-cli/
Run Devin on one branch and Codex on another. Same repo, isolated worktrees, no collisions. Keep the better diff.
That is the point of a multiplayer harness: the harness is a choice, not a commitment.
Devin's pitch is start local, hand off to the cloud.
We give the local half a permanent home. The session runs on a machine that does not sleep. Close the laptop, open it from your phone, it is where you left it.
Connect it once in Connections. Sign in with your own Devin account. Every workspace can start a Devin session the way it starts Claude Code or Codex.
Your Devin plan pays for Devin. AQ adds no markup.
Devin CLI @cognition now runs in AQ.
Cognition's terminal agent, on an always-on machine, in its own git worktree, with your team watching the same live session.
AQ retweeted
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.
