Smart model routing is the quiet infrastructure shift of 2026

Gergely Orosz just published a piece on The Pragmatic Engineer's blog about how teams are routing AI requests across models instead of hardcoding everything to one. The idea is simple: send cheap queries to a cheap model and big jobs to a beefy one, and let the routing layer decide which is which. It's not a flashy product — it's plumbing — and that's exactly why it matters.

The pieces already exist. OpenAI, Anthropic, Google, and the open models all ship APIs that accept the same input and return the same output. The routing layer sits in the middle, watches latency and cost, and picks the best model for the job. No vendor lock-in. No rewriting your app.

The reason this is worth writing about is timing. Two years ago, routing was a novelty — you paid a premium for the smartest model and didn't think twice. Now, the cost curve has flipped. Running a single model for everything gets expensive fast. Routing is how teams keep bills flat while the model quality stays up. The writers at TLDR call it a new trend for 2026, and the Pragmatic Engineer's note is the clearest explanation of why: it's infrastructure that scales without you noticing.

Why this matters for us: the abuelas running side businesses don't need the fanciest model — they need reliable tools that don't cost a fortune. Smart routing is how that happens.