Use case · Route by difficulty

Every request pays frontier prices. Most don't need to.

A routing layer in front of your pipeline sends each request to the cheapest model that can handle it — a fine-tuned small model in your infrastructure for the routine share, your existing frontier API for the genuinely hard cases. Nothing gets dumber; the bill gets smaller.

Model routing places a lightweight decision layer in front of your AI pipeline that sends each request to the cheapest model capable of handling it — a fine-tuned small model in your infrastructure for the routine share, your frontier API for the hard cases. When roughly three-quarters of traffic routes local, blended cost falls around 70% (our estimate from published list prices, August 2026) — and frontier capability stays in the loop.

The problem isn't which model you use. It's that you're only using one.

One-model pipelines charge you the same rate for “classify this email” as for “reason through this edge case” — difficulty varies per request, price doesn't. But ripping out the frontier model entirely is the wrong fix: some fraction of your traffic genuinely needs it.

How it works · The Crit Path
01Analyze — score the trafficWe sample your real requests and score them by task type and difficulty. In most production pipelines, 60–85% of requests are routine enough for a small model — we establish your actual number before promising anything.
02Triage — build the routing layerHitbox, our eval-and-routing harness, sits in front of your pipeline: a lightweight classifier plus confidence scoring decides, per request, whether the local model answers or the task escalates to your frontier API. The router itself adds milliseconds, not a second model bill.
03Fine-tune — the local workhorseA 3–7B open-weights model, trained on the routine slice of your traffic and deployed in your VPC. The frontier API stays in the loop for the hard tail — quality is preserved by design, not by hope.
04Measure — then tightenEscalation rate is a tracked metric, not a mystery. As the small model improves through re-training on escalated cases, the routine share it handles grows — and the blended cost keeps falling.
The economics

Illustrative math, from public list prices, August 2026. Take contract-page review at roughly $0.30 per document on a frontier API. Route 75–80% of documents to a local 7B model running at a fraction of the per-token cost, escalate the rest, and the blended cost lands near $0.05 per document — about 70% off, with the hard cases still going to the big model. The arithmetic generalizes: blended savings ≈ routine share × small-model discount. Our cost model; The Crit replaces it with your numbers.

One frontier modelRouted: SLM + escalation
Price per requestFrontier rate on every call, easy or hardRoutine share at small-model rates; hard tail at frontier
Worked example: document review$0.30 per document (public list prices, Aug 2026)~$0.05 per document blended — our estimate
The hard casesHandled — and priced into every routine call tooEscalate to your frontier API by design — no quality cliff
Quality driftInvisible until the complaints arriveEscalation rate: a tracked, alertable metric

For the underlying cost evidence on small-model inference, see the published migrations on our LLM-to-SLM Migration page (Forethought, 66–80% lower inference cost, vendor-reported, on AWS).

Where this doesn't work

Savings scale with your routine share. If less than half your traffic is routine — or your volume is small — a router won't pay for its own build cost. We measure the share first, in The Crit, and give you the number either way.

What you get

The triage layer. Yours.

  • Traffic analysis: difficulty distribution and your measured routine share
  • Hitbox routing service deployed in your infrastructure, with per-request decision logs
  • A fine-tuned 3–7B local model for the routine slice — weights are yours
  • Escalation policy tuned to your accuracy bar, with a frozen eval set gating it
  • Dashboard: cost per request, escalation rate, accuracy by route — the three numbers that matter
Who this is for
Teams that can't leave frontier quality behindBut shouldn't pay for it on every call — routing keeps the capability and drops the blanket price.
High-volume pipelines with mixed difficultySupport triage, document review, agent loops — where easy and hard requests share one queue.
Builders of agentic systemsTool-calling steps burn tokens on trivially structured work; a router sends them down the cheap path.
Questions, answered straight

What is model routing (LLM routing)?

Model routing places a lightweight decision layer in front of an AI pipeline that sends each request to the cheapest model capable of handling it — typically a small fine-tuned model running locally for routine requests, with escalation to a frontier API for hard ones. It cuts cost without removing frontier capability from the system.

How much does a hybrid SLM/LLM router save?

Around 70% blended, when roughly three-quarters of traffic routes to the small model — the arithmetic is routine share times the small model's per-token discount, which runs 10–100× on published list prices (August 2026). The savings depend entirely on your routine share, which we measure from real traffic before the build.

Does routing to a smaller model reduce output quality?

No — that's the point of the escalation path. The small model only answers requests it handles at or above your accuracy bar, verified on a frozen eval set; anything below the confidence threshold escalates to the frontier model automatically. Escalation rate is monitored in production, so quality drift shows up in a metric, not in complaints.

Security · by architecture

Cut the bill. Keep the custody.

  • Data stays in your VPC
  • No new subprocessor
  • Escalation under your rules
  • Keys and logs are yours
Architected to deploy inside your
SOC 2 GDPR ISO 27001
Small models · critical hits

Stop paying reasoning rates for routine work.

Two weeks of traffic analysis tells you your routine share, your escalation rate, and your blended cost — before any build starts. The numbers are yours either way.