Story · Arize
How I cut coding agent costs with model and harness routing (Arize)
blog post · Story page
An author writing on Arize routes the phases of a recurring coding-agent workflow to different models and harnesses and reports a large per-run cost drop.
In plain words
- An engineer reports cutting the cost of a repeated task that uses artificial intelligence to write computer programs.
- Different systems handle planning the work, exploring the code, writing changes, and checking the results.
- The engineer says the change brought their cost per run from roughly $100 to $15-$20.
Appeared in
- Per-phase model routing cuts agent cost, RASER on Slurm, Amp steers mid-run
Sep 09, 2026 · lead story
Subscribe
Get the brief in your inbox
Pick daily, weekly, or both. Nothing is gated either way: every issue is on the site and in the feeds.
- Weekdays at 8:45am IST, one lead story and 6 to 9 items.
- Sundays, an argued synthesis rather than a recap.
- One click to leave, and quiet days say so in the subject line.