Daniel Antonio

what are some solutions in opencode to create `agent-teams` like interactions? thinking to orchestrate using mimo v2.5 pro and running the executions using cheaper models with less context limit.

playing around with @opencode and being pleasantly surprised with the quality i get from open weight models, especially @mimo_labs V2.5 pro. fast, light, 1m context, intelligent enough for my current use case. note: starting a new project, will update on my experience again

@TheAhmadOsman getting the 24gb one. will save up for GPUs instead haha

@TheAhmadOsman hey Ahmad, while it’s not multi GPUs. im thinking of buying a new mac and running some local LLMs. not sure if it’s better to get a normal mac (m4 pro 24gb) and get GPUs later, or get a better mac w/ more ram (48gb?) to run them. your input would be greatly appreciated :)

Three opensource models were announced while I am on the same flight and you're still not bullish on opensource AI, anon?

@moofeez coming from a web SE, what do i need to know to be able to do this?

I post-trained Qwen3-Coder to fix bugs using an actual debugger. The result: Solve rate: 70% → 89% Median turns to fix: 46 → 19 (-59%) Instead of just reading code or print-debugging, it: - reasons from execution - inspects live variables and call stacks - sets breakpoints, steps, and evaluates expressions

@conductor_build @charlieholtz is there a way to configure codex usage using my config.toml file? not sure how to configure bedrock open ai models

Yes

@charlieholtz any chance conductor.build could push a terms of service page? i’m pushing the adoption of @conductor_build in my company but blocked due to it not having terms of service

We've just fixed a bunch of rendering and performance issues. Idle CPU usage is much lower now and the app should feel snappier!