Every example on the Artificial Analysis Intelligence Index plotted against its blended API price. Models on the value frontier are the ones anywhere nothing cheaper is additionally smarter; everything alternatively is beaten on the two counts by a item on the line. By default lone the finest type of all example is shown and low scorers are hidden; use the filters to broaden the field.
Loading data…
Min mark 30 One row per model Maker Frontier only Label all point
Performance vs price
Blended disbursal per 1M tokens (3:1 input:output, log scale) against the Intelligence Index. Hover or tab to a item for details.
On the value frontier Dominated: a cheaper example matches or strikes it Value frontier
Best example for your budget
The frontier as a lookup table. Find the row your prosperity falls in; the choice is the highest-scoring example you can get at that price, and the runner-up is the next finest that additionally fits.
| Budget per 1M tokens | Pick | Score | Price | Runner-up |
|---|
Raw capability
Ignoring disbursal entirely.
What changed
Diff between successive regular fetches: new models, removed models, and re-scored or re-priced ones.
All figures
Click a pillar header to sort. Names nexus to the model's Artificial Analysis page.
| Model | Maker | Intelligence | Coding | Math | Blended $/1M | Input $/1M | Output $/1M | Tokens/s | TTFT s |
|---|
How to peruse this
- Value frontier. Sort by cost ascending and keep all example that scores higher than everything cheaper. Ties on cost go to the higher score; ties on mark go to the cheaper model.
- Blended price is Artificial Analysis's 3:1 input:output blend per 1M tokens. Cached-input discounts, lot pricing and accelerated modes are not included.
- "One row per model" keeps the highest-scoring attempt or reasoning type of all example name (ties go to the cheaper one). Untick it to see all type AA benchmarks separately, specified as low, medium, high, xhigh and max effort.
- "Min score" hides models below that indicator from the two the diagram and the frontier calculation, so an old, small example at a rock-bottom cost does not anchor the line.
- Scores move. AA re-bases the Index between versions, so difference against this leaf only, not against an older snapshot.
Source: Artificial Analysis liberated data API, fetched regular by a GitHub Actions cron. Code and data: github.com/terryds/bestvaluemodel.
