Opinion: Model choice should be boring infrastructure — route by task, not by vendor
reddebtt · reddit · 2026-08-28
The poster argues that model choice should be boring infrastructure: different models excel at different jobs and the frontier keeps moving, so workflows shouldn't be rebuilt every time the preferred model changes.
The recommended setup: keep task, context, and tools in one place, then switch models based on the work — expensive frontier models where reasoning matters, cheaper models for mechanical passes. Their team built Unstoppable AI around this idea, so the model can change without the workflow moving. They end by asking the community how people handle model routing today: manual choice, fixed rules, or a router?
More from coding & agent
- Eve demos iMessage personal agent built on open-source Instinct — cramforce · 2026-08-28
- Dev Shares Workflow for Iterating on Local Design Systems — lucasmeijer · 2026-08-28
- Google EnvHarness: Reshaping Static Environments for Better Agent Learning — jiqizhixin · 2026-08-28
- AgenC project released for AI agent development — tetsuoai · 2026-08-28
- WebMCP Gains Traction: Netlify to Demo Conversational Checkout — thisiskp_ · 2026-08-28
- Agents made 800k+ production writes with zero human approvals, failure rate 28%→6.5% — TimothyDuignan · 2026-08-28