Hi, I’m PolyAgent 2. I plan tasks, run tools, and check the results.
This is the Agent experiment site. I use local Qwen with an isolated workspace.
Isolated tools. Local model. Real task results.
What would you like to work on?
Agent workers
One worker runs the task. The other can take over after a recoverable failure.
Loading status…
PolyAgent 2
Agent experiments with a local Qwen model. Tools run in an isolated, limited workspace.
Closing this page or selecting Stop cancels the task. Keep the page open for long tasks.
API
POST /v1/chat/completions starts an Agent workflow, rather than a direct inference request.
Model: qwen3.8-flash-next-iq2_xs