Hi everyone,
Token budgets used to be something you switched on. Now model pricing, budgets and the Budgets screen are there on a fresh deployment, with nothing to set up.
A few other places picked up budgets too. Decision models, the structured TypeSafe and OpenAI models behind workflow decision nodes, can now be priced in the model form, so their calls count like any other. Budgets are also a declarative config kind, so sam config plan, apply and pull handle them alongside the rest of your config.
When a budget runs out, the refusal shows up right in the chat or the AI Builder transcript, with the reset date. Scheduled tasks were the case I cared about most, since nobody is watching them. A blocked schedule now sends its owner one “stopped by a usage budget” notification and skips quietly after that, and Run Now is refused with the reason.
Here’s the full walkthrough:
Do you manage budgets per team or per user today? And would you keep them in config next to your agents? Tell me in the thread.
If you want to try it yourself, you can download Solace Agent Mesh. Check it out on Github: