We just shipped something that we think is a glimpse into the future: MAX Agent.
It’s available now for Silver and Gold members. You’ll find it in the dropdown next to your chat input.
What’s Different?
Max Agent is greedy. It’s optimizing for maximum intelligence at the expense of speed and cost. It leverages Claude’s most powerful model (Opus 4.5, an amazing model) to do the heavy lifting. It’s smarter, more capable, and better at reasoning through complex problems.
Where it really shines:
- Complex third-party integrations - APIs, webhooks, external services
- Sophisticated designs - layouts and UIs that require more nuance
- Advanced business logic - role-based access control, admin systems, multi-step workflows
- Hard debugging - when you’re stuck on something tricky
It’s noticeably better at code in general. If you’ve hit a wall with a complex feature, Max Agent is worth trying.
The Tradeoff
Max Agent uses more credits per request. That’s the honest reality - smarter models cost more to run.
But here’s the thing: it also makes fewer mistakes. For complex tasks, you might actually save credits over a longer session because you’re not going back and forth fixing issues.
We recommend trying it when you’re working on something genuinely difficult, not for quick tweaks.
Pro Tip
Stick with one agent per session. Mocha uses aggressive caching to save you credits on follow-up requests. Switching between Max Agent and the default agent resets that cache, so you’ll spend more than necessary.
Try It Out
If you’re Silver or Gold, it’s already live in your account. Give it your hardest problem and see what happens.
We’d love to hear how it goes - drop your feedback in our Discord.
Build away ☕!