Intelligence ranking
Top 18 by intelligence score.
Loading pricing and benchmark data…
Compare leading AI models by intelligence, cost and performance. Every number carries its source and the date it was checked.
Every model with a verified price, ranked. Sort by any column, filter and export. Click a row to open the full record with every source.
How this ranking is built →| Model | Provider | Input / M | Output / M | Intelligence | Context | Value for money | Monthly cost | Source |
|---|
Three questions, in plain language. The estimate uses the same official prices as the table above and updates as you type. Not sure how big your work is? The examples above load straight into it.
These three numbers are assumptions made by Avital, not provider data. Replace them with your own and everything recalculates.
Set a spending ceiling and a quality floor. Anything over budget is dropped first, then what remains is ranked by intelligence.
The migration question: how much you save, and how much capability you give up.
The same data seen by capability, by efficiency, by spend and by provider.
How this ranking is built →The horizontal axis is logarithmic: every step to the right multiplies the cost. Up and to the left buys more capability per dollar.
Top 18 by intelligence score.
Intelligence points bought per dollar. Logarithmic scale: the gaps are orders of magnitude.
The 18 cheapest for the parameters in the bar above.
How many models sit in each price band.
Average intelligence against average blended cost. Only models with a verified price are included.
What we could verify, what we could not, and the exact formulas. A missing number is more useful than an invented one.
Five questions, answered before you have to ask them: who gets in, where the score comes from, how the order is decided, what we never do, and what this ranking cannot tell you.
Prices come from each provider’s own published pricing page. Where that was not reachable, from an aggregator that cites the official source — always labelled as such.
| Source | Type | Dates |
|---|
Excluded from every figure: batch discounts (−50% at almost every provider), cache hits (up to −90%), and surcharges for long context, data residency or priority mode. Each of those moves your bill more than the gap between two neighbouring models in this table.