Why Codex has become my everyday AI tool
My current preference is less about collecting models and more about having one tool that fits my agent-driven working day.
Expanded from a LinkedIn post originally published on 2026-10-06.
I have not used any Anthropic models in about 3 months.
Codex is cheaper, I can use it with Hermes and the new Astra-6.1 Sol model has literally changed the game as a 1 size fits all model for all my needs.
I pay £80 a month and never hit my limits. If I do I get free resets I can use. Sometimes OpenAI give them to me as a surprise. I’ll log on in the morning and my usage is at 0% again.
Tasty!
#agenticcoder #agenticlife
The useful tool is the one that fits the day
I have spent the last few months using Codex rather than switching between Anthropic models. For my current workflow, the combination of cost, capacity and being able to use it with Hermes has mattered more than maintaining a collection of competing tools.
The price and usage experience in the original post describe my own account at the time. They are not a universal pricing comparison or a guarantee that someone else will have the same limits or resets.
Less switching, more working
A model that works across the tasks I actually give it is valuable because I spend less time deciding which tool to open. I can put that attention into the brief, the implementation and the review instead.
That is a preference, not a declaration that one provider wins every possible task. For me, right now, a dependable everyday setup is more useful than continually rebuilding the workflow around the latest comparison.