Skip to content
22decisions home

Most Of Your AI Doesn't Need A Frontier Model

Open Models, Frontier Models, And The Case For Two-Tier AI

Published

Most enterprises buy AI as if every task needs the most powerful model available. Most tasks do not. Open models now carry most developer AI traffic at about a quarter of the price, while frontier models keep their edge on hard, high-stakes work. Technology teams have moved, in India as much as the West, but business leaders may not know. This Provocation makes the case for two-tier AI and gives a 30-day plan to route each task to the right model.

Blended research: grounded in surveys, powered by AI (read our methodology)

Key statistics

  • 42%

    of enterprises run open-weight models in production, up from 31% two months earlier

    Evidence: Cited

    Futurum, September 2026

  • About 1/4

    the token-weighted list price of open models compared with closed models

    Evidence: Cited

    OpenRouter, October 2026

  • 56% / 14%

    open-weight models' share of tokens and of estimated spending on Vercel's AI Gateway

    Evidence: Cited

    Vercel data, as reported by Futurum

Download this Provocation free

You'll also receive our monthly briefing. Unsubscribe at any time.

Trouble downloading? Email hello@22decisions.com

Summary

Summarising, classifying, extracting and drafting make up much of enterprise AI volume, and open models now handle much of that work at a fraction of the price. Closed models still lead on the hardest tasks, but for routine work the gap matters less than the price.

This sixth issue of 22decisions Provocations compares open and frontier models on price, capability and adoption, shows how far technology teams have already moved, in India as much as in the West, and sets out why most enterprises cannot simply switch.

Our provocation

Paying frontier prices for routine work is the AI equivalent of sending every letter by courier. The aim is not one model, but the right model for each task, with a provenance policy that business leaders, not just engineers, have agreed.

What's inside

  • What leaders believe, and what the evidence suggests
  • The price gap is large; the capability gap is narrower
  • Technology teams have moved; leadership may not know
  • Why most enterprises can't simply switch
  • Tier your AI in 30 days

5 pages, PDF. Free to share with attribution.

Written byVarun Shourie, Founder, 22decisions

Varun Shourie founded 22decisions to enable technology leaders to prepare for, and get ahead in, the AI era.

About 22decisions

Download this Provocation free

You'll also receive our monthly briefing. Unsubscribe at any time.

Trouble downloading? Email hello@22decisions.com

Topics:
AI costs, Open models, Model strategy, AI leadership
Sectors:
Cross-industry

Get our research in your inbox

One considered briefing a month on how organisations decide on, buy and adopt AI. Unsubscribe at any time.