Opus is an Anthropic/Claude model, not an OAI/Codex model. I’m not sure if you can actually use an Anth model on Codex these days, but my experience to date indicates that performance tends to be best when you match the model to the harness it’s intended for.
You might also consider using the mid tier model (opus, Terra) to generate a detailed plan and then use a lower tier model (sonnet, Luna) to actually implement each task in the plan.
I’ve gotta say my experience has been the opposite. I’ve been consistently surprised at how much mileage I can get out of the $20 plan.
My usage is pretty simple. I just use codex with an AGENTS.md to point it in the right direction. I have no MCPs and use Sol on medium thinking as the main model.
For longer sessions I tell it to launch a sub agent to make changes. Otherwise I just reset my context for each new task.
Start auto-compacting your context at 100k. Your sessions will incur exponential token usage if you don't draw the line before the 1m context limit. It's the best way to efficiently use a small token quota for long-horizon tasks.
Opus is an Anthropic/Claude model, not an OAI/Codex model. I’m not sure if you can actually use an Anth model on Codex these days, but my experience to date indicates that performance tends to be best when you match the model to the harness it’s intended for.
You might also consider using the mid tier model (opus, Terra) to generate a detailed plan and then use a lower tier model (sonnet, Luna) to actually implement each task in the plan.
I’ve gotta say my experience has been the opposite. I’ve been consistently surprised at how much mileage I can get out of the $20 plan.
My usage is pretty simple. I just use codex with an AGENTS.md to point it in the right direction. I have no MCPs and use Sol on medium thinking as the main model.
For longer sessions I tell it to launch a sub agent to make changes. Otherwise I just reset my context for each new task.
OpenCode and cheaper plans await if you can move beyond the Big Ai labs and their bluster
Start auto-compacting your context at 100k. Your sessions will incur exponential token usage if you don't draw the line before the 1m context limit. It's the best way to efficiently use a small token quota for long-horizon tasks.
Um, you can put in $20 for more tokens any time.
[dead]