Optimus-1 is now generally available in ArkenOS, the desktop research workspace from Iacon Autonomics, a research lab in Palo Alto. A team sets the problem, and Optimus-1 starts agents that search hypotheses in parallel, follow outliers, and keep what replicates until a branch becomes an artifact or is closed. The release is for reinforcement-learning labs, data teams, and longer investigations, where most branches fail and the hard part is knowing which result to believe. The workspace runs on the team's own hardware, with unlimited agent concurrency, and teams connect their own model keys or supported subscriptions. Paid access starts at $200 per user per month for up to three people, and at $2,500 per month for teams of up to ten, with a 30-day trial on the team's own provider account. The design target is a long investigation that continues after failed runs, keeping the results that replicate. Dev Shah, who is building Optimus at Iacon and is head of developer relations at Resemble AI, announced the release. Iacon has already used Optimus with Resemble AI to build DetectBench, which scores investigative agents on deepfake cases and records the tools and evidence behind each decision. Separate writeups cover a Gemma painting pipeline with Google DeepMind and a MuJoCo workflow for a Franka Panda picking task. Shah says ArkenOS is already helping hundreds of teams, from RL environments and data labs to frontier research groups.