Dev workflow: iterate plans with big models, then hand off execution to small reasoning models
intellectronica · x · 2026-09-05
Developer electronica shares his model-usage pattern: he does most of his work with small reasoning models (luna, gemini, glm, qwen, deepseek flash), reserving large models for open-ended interactive work and serious analysis/planning.
His emerging workflow: "vibe with the big models" to iterate on plans, then ship those plans to small thinking models for background execution — balancing intelligence with cost and speed.
More from coding & agent
- Dev says GPT-6 Astra solved his months-stuck mocap retargeting workflow in two hours — tinyfool · 2026-09-05
- Open-source AI trading agent turns social sentiment into signals via Gemini — tom_doerr · 2026-09-05
- A dev building an Epub plugin breaks down the evolution of TTF, OTF and WOFF2 — vista8 · 2026-09-05
- Codex community plans ~40 global meetups with hackathons and multi-agent workshops — gabrielchua · 2026-09-05
- Magnitude: open-source server picks the best local models for your hardware and plugs into your coding agent — solyarisoftware · 2026-09-05
- Dev patches WebKitGTK CVE to restore drag-and-drop, ships Copilot app as Flatpak — unixterminal · 2026-09-05