What model is the best at finding vulnerabilities inside agents?
👽 Enoki Labs's attacker runs on a harness plus a model. The harness is ours. We put 10 models through it, against two agents with 23 known vulnerabilities. One writes code, one moves money. Only the model
DeepSeek V4.1 Flash Dropped the sausage!⚡⚡⚡
Today is September 22nd, the crazy bot dropped the sausage. Some kind of f*ck! Call Mikhail Svetov! What is the bot doing? Why the hell did you drop your sausage?! bot response “I dropped it because in the model the friction force between the fingers and the slippery
Someone created an 'OpenRouter' for AI agents.
Call it HarnessRouter: a unique interface that runs Codex, Hermes, Claude Code, DeepSeek Harness and 9+ others. Anyone who switches between these tools on a daily basis has everything in one place. Which one would you try first?
The more I use MossAI, the more I start looking at AI access differently.
We’re getting used to paying for AI every month. Subscriptions, API costs, credits… you pay for the access, use the model, and move on. But Model Max made me stop and think about that relationship a
You have reached the end of the archive
All of deepseek