All updates
v2.3
Agents
Model routing per agent
Pick a different model for each agent, or let Orrery route every request to the fastest model that can answer it well.
Each agent can now run on its own model. Scribe can stay fast while Analyst uses a larger model for long research questions.
What’s new
— Per-agent model settings in workspace admin
— Automatic routing that balances speed, cost and quality
— Usage broken down by agent and model