Artificial Analysis has built a ranking of language models, updated daily from free data, that tells you which one is the best value for your money. The site plots every model on its Intelligence Index against its blended API price, and the result is a simple frontier: the models where nothing cheaper is also smarter. Everything else gets beaten on both counts by a point on the line.
The ranking is a lookup table. You find the row your budget falls in, and the pick is the highest-scoring model you can get at that price. The runner-up is the next best that also fits. The whole thing is fetched daily by a GitHub Actions cron from the Artificial Analysis free data API.
How the Frontier Works
The chart shows blended cost per 1M tokens on a 3:1 input:output log scale, against the Intelligence Index. Hover over a point or tab to it for details. By default, only the best variant of each model is shown, and low scorers are hidden. Filters let you widen the field.
Sorting by any column header moves the order around. Names link to the model’s Artificial Analysis page.
The Value Tradeoff
The value frontier is where the tradeoff between cost and intelligence stops being painful. The ranking is designed to show you where cheaper models are also dumber, and where smarter models cost more. The frontier itself is the point where neither condition holds — nothing cheaper is also smarter, and nothing smarter is also cheaper.
This is the kind of tool that makes technology feel less like magic and more like a shopping list. It takes the guesswork out of comparing models across price and performance, and it updates automatically so the answers stay current.
Daily Fetches and the Data Pipeline
The site is updated daily through a GitHub Actions cron job. That means the data comes from fresh fetches, not a static snapshot. New models appear, removed models disappear, and existing models get re-scored or re-priced. The diff between consecutive daily fetches tracks those changes.
The code and data live on GitHub at github.com/terryds/bestvaluemodel. Anyone who wants to see how the rankings are generated can look at the source.
Ignoring Cost Altogether
There is also an option to ignore cost entirely. That lets you see the models ranked purely by intelligence, without the price filter getting in the way.
For everyone else, the value frontier is the practical answer. It tells you which model delivers the most intelligence per dollar, and it does it without requiring a degree in machine learning.
What We Make of It
The ranking is a small trick, but it is a useful one. It takes two things — a measure of a model’s intelligence and its price — and puts them side by side so you can compare them directly. The daily update means the information stays fresh, and the free data API means anyone can build on top of it.
This is the kind of tool that belongs in a developer’s toolkit. It saves time spent hunting for benchmarks and pricing tables, and it provides a common ground for comparing models that might otherwise be hard to judge against each other.
Key Facts Box
- Ranking: Best LLM for every budget, updated daily
- Data source: Artificial Analysis free data API
- Fetched: Daily via GitHub Actions cron
- Code/data: github.com/terryds/bestvaluemodel
- Plotted: Blended cost per 1M tokens vs Intelligence Index
Schedule Table
| Action | Frequency |
|---|---|
| Daily fetch of model data | Every day |
The ranking is a reminder that technology does not have to be opaque. A model’s intelligence has a number, and its price has a number, and putting them together produces a decision. That is a small victory, but it is a real one.
Source material: “Best LLM for every budget, updated daily,” terrydjony.com.
Get the Notebook.
The day's best stories and every fresh verdict, in plain English, in your inbox by seven. One email a day, no more.