AI
Liner Just Shipped an API That Picks the Cheapest Model for Every Single Prompt
Here is the enterprise AI habit that stays out of the slide decks. Companies pay frontier model prices for work a much smaller model could handle, then act surprised at the invoice. Liner, an AI agent solutions company, launched the Liner Model API on September 24 to kill that habit at the plumbing level. The API routes each request to the most cost efficient model capable of handling it, automatically, with the API choosing instead of manual benchmarking or model switching by the developer.
The mechanism is refreshingly boring, which is the compliment. The Liner Orchestrator evaluates the expected quality and cost of candidate models for each request, then picks one, a single model per call rather than fanning out to several at once. That single model choice matters because multi model approaches stack token costs with every extra call. One API now covers the spread from everyday questions to coding, reasoning, and deep research workloads that used to demand ongoing model selection and cost tuning by hand. The routing decision happens per request, in real time, while the developer keeps a single endpoint and the API does the choosing behind it.
Liner has receipts from its own kitchen. After deploying the Orchestrator, the company's internal token expenses in August fell more than fifty percent compared with the first half of 2026, following performance validation in real world usage plus controlled benchmark testing. A number like that lands because it comes from production traffic rather than a slide deck. When the vendor eats its own routing, the pitch stops being theory.
The bigger story is what this says about where the AI stack is maturing. The model layer is racing to the bottom on price, as this week's launches showed, and the routing layer is where the savings get captured. The winners in enterprise AI may be the companies that skip training entirely and simply get very good at spending other people's intelligence wisely. Routing sounds unglamorous next to a new frontier release. It is also the line item the CFO actually reads, and the one most likely to survive the next budget review.
If you ship AI features, the practical move is an audit of your own traffic. Count how many of your requests truly need the flagship and how many are paying flagship prices out of habit. Tools like Liner's make the answer automatic, but even a manual pass usually finds the same shape. Most of the tokens were rarely the hard part. Picking the right brain for each question was.
Quick answers
What is this story about?
Here is the enterprise AI habit that stays out of the slide decks. Companies pay frontier model prices for work a much smaller model could handle, then act surprised at the invoice. Liner, an AI agent solutions company, launched the Liner Model API on September 24 to kill that habit at the plumbing level. The API routes each request to the most cost efficient model capable of handling it, automatically, with the API choosing instead of manual benchmarking or model switching by the developer.
Why does this story matter?
If you ship AI features, the practical move is an audit of your own traffic. Count how many of your requests truly need the flagship and how many are paying flagship prices out of habit. Tools like Liner's make the answer automatic, but even a manual pass usually finds the same shape. Most of the tokens were rarely the hard part. Picking the right brain for each question was.
Sources
New to crypto? Read the crypto glossary, browse frequent questions, read our story, or explore the story archive.