GPT-5 was the best model in the world a year ago. Today it's inferior to Qwen3.6-27B, an open weight model you can run on your laptop.
Progress is brutal.
If you're wondering why GPT-5.6 Sol with subagents in the Codex CLI burns through your usage so fast, there seems to be a bug. 👀
It can't spawn smaller subagents for basic exploration or web research and forces Sol for everything. Which is annoying because as with Fable, I'd love the Advisor / Executor workflow.
Anyone else seeing this too?
Workaround for now: add this to your ~/.codex/config.toml:
[features.multi_agent_v2]
hide_spawn_agent_metadata = false
tool_namespace = "agents"
I've been using Grok Build some time now, and it has genuinely surprised me. It is unique in the current agents landscape, has a different tone, a different feel.
It can search the web and X very fast and is improving even faster every day.
Give it a try and let me know if you think the same!