After careful consideration, we’ve made the decision to wind down @code4rena. This community has meant a great deal to everyone who has been part of building it, and sharing this news is not easy.
@0xLevente You can set a limit by typing “budget X tokens”, but I’ve never experienced a major drain without a budget
Double check if “Fast mode” is enabled, it is enabled by default now and it literally sucks up the 5h limit in minutes
One of the biggest advancements in AI have been Ralph loops, now called `goals` in Codex
Goals allow your agent to literally work for hours or until the mission is complete. BUT, as always, you need a good prompt to guide it and validate its progress. Plan mode or any plan related skills can help with this.
If you're not seeing it in your Codex CLI, update it to 0.128.0+ and run `codex features enable goals`.
I've been using Codex exclusively for months now, love it.
Never liked Claude pushing people to the $100 plan, so that's out of the picture.
What's the next best coding agent/model?
@oakenknight19 Glad to hear that!
I've always preferred Codex, although people are recommending more underground agents rn, like Qwen, Hermes and others, so I'm looking for backup
@levikov Serbian here, basically nobody has worked for less than $1000-1500 for years now (and that’s net), so this “alpha” is kinda outdated. Also, many are moving to direct B2B now, instead of agency work where you get scalped by your employer (myself included)