Token Budget Engineering
First time hitting my $200 Claude Max plan limit - seems like we're entering an era where demand for tokens will far outpace supply
I've been on the $200 max plan since July of last year, this is the first time I've hit a limit.
To be fair, I am continuing to push the boundaries through things like repeating automated review + fix cycles, and it's clear Anthropic is having capacity issues.
Good thing I was already in the process of adapting my skills to codex, and I've got ollama set up to dabble with local models.
Seems like we're entering an era where demand for tokens is going to be way bigger than the supply thanks to agents being really useful. Not just context engineering, but token budget engineering too.
For me, it's just as much about the workflow / whole SDLC as it is the model.
I'm not upset I splurged to get 128GB of ram on this new laptop to see what I can offload to local models. I don't think it can keep up with my 5-10 parallel sessions but if things keep clamping down... life, uh.... finds a way 🦖
Related insights
Codex Is Becoming My Daily Driver
After a year with Claude Code as my daily driver, the Codex desktop app's interface for managing many agent threads and automations has me reaching for it more and more.
What If Money Was No Object?
A CFO asked me that while I was on the hot seat. Agents change the old Mythical Man-Month math—but going fast in the wrong direction is still waste.
You Don't Need Opus for Everything
Right model, right task. In Claude Code you can pin a model per skill—I drop to Sonnet or Haiku when the full power of Opus isn't needed to stay inside my limits.
Get More Like This
Follow along as I build and share what I learn
Found this helpful? Share it with your network!