After burning 1.5B Astra tokens: 4 practical Codex tips
brandon_galang · x · 2026-09-08
After burning 1.5 billion GPT 6 Astra tokens over a weekend, the author shares 4 Codex tips:
- Start on low effort: Astra low outperforms 5.6 sol high; step up to medium/high only when deeper reasoning is needed.
- No cache break on effort switches: unlike most models, switching reasoning levels doesn't invalidate your cache—crank up for planning turns, drop back to low once certain.
- Raise context limits: Astra doesn't charge extra for long context in Codex; the author runs 400k context window with a 360k compact limit.
- Give it tools and let it run: Astra shines at computer use; pair with the AsideAI browser REPL and it can one-shot end-to-end tasks across apps.
Verdict: use Astra for ambitious, general agent workflows.
More from coding & agent
- VCs debate whether training-coupled harnesses will beat generic harness plus model — ruslansv · 2026-09-08
- Give a spreadsheet agent a response budget, not just a cell-range argument — Suspicious_Shift8979 · 2026-09-08
- Microsoft open-sources tgrep, a trigram-indexed grep up to 52x faster than ripgrep — leslysandra · 2026-09-08
- Vercel engineer builds a new syntax highlighter, benchmarks correctness against Shiki — shuding · 2026-09-08
- One screenshot, one prompt: agent builds USS Enterprise 3D model in Blender for ~15 EUR — rschu · 2026-09-08
- ML syntax highlighter matches Shiki accuracy with the model inlined in the bundle — shuding · 2026-09-08