How much intelligence does a blended dollar buy?
Intelligence has a price, and the price is convex. This page divides each frontier model's Artificial Analysis Intelligence Index score by its blended API price, weighting input 80% and output 20%. The last three index points cost more than triple the money. Click any column to sort.
| MiniMax M3 | 44 | 0.30 | 1.20 | 0.48 | 91.7 |
| Muse Spark 1.1 | 51 | 1.25 | 4.25 | 1.85 | 27.6 |
| GLM-5.2 | 51 | 1.40 | 4.40 | 2.00 | 25.5 |
| Gemini 3.5 Flash | 50 | 1.50 | 9.00 | 3.00 | 16.7 |
| Qwen3.7 Max | 46 | 2.50 | 7.50 | 3.50 | 13.1 |
| Kimi K3 | 57 | 3.00 | 15.00 | 5.40 | 10.6 |
| Claude Opus 4.8 | 56 | 5.00 | 25.00 | 9.00 | 6.2 |
| GPT-5.6 Sol | 59 | 5.00 | 30.00 | 10.00 | 5.9 |
| GPT-5.5 | 55 | 5.00 | 30.00 | 10.00 | 5.5 |
| Claude Fable 5 | 60 | 10.00 | 50.00 | 18.00 | 3.3 |
Blended cost per 1M tokens = 0.8 × input price + 0.2 × output price, the mix of a retrieval-heavy workload that reads far more than it writes. Scores are the Artificial Analysis Intelligence Index at each model's highest stated reasoning effort; prices are first-party API list prices. The ratio rewards cheapness, so compare models within a capability tier rather than across tiers · a model that fails the task is expensive at any price.
GLM-5.2 and MiniMax M3 are open-weight models; the rest are closed. Moonshot has promised Kimi K3's weights for 27 Jul 2026.
Source: artificialanalysis.ai per-model pages, retrieved 16 Jul 2026. Prices and scores move; the retrieval date is the claim.