Claude beats ChatGPT
The $30 AI model isn't #1 — one benchmark, five models, one surprising gap.
Five models plotted by standard API output price against Artificial Analysis's Intelligence Index v4.1.1. This is one benchmark snapshot, not a universal ranking.
GPT-5.6 Sol costs $30 per million output tokens. Claude Opus 5 costs $25 — and scores 2 points higher on this index.
Muse Spark 1.2 scores 54 at $4.25/M output. DeepSeek V4 Pro scores 53; its $3.96 figure is the official peak-hour output price. Gemini 3.1 Pro Preview scores 48 at $12/M output.
Prices are output-token list prices, not the total cost of completing a task. Reasoning effort, token use, caching and retries can change the bill.
One index, one date. Not financial advice.
Next: what an AI chip actually costs to build, and who gets paid for each part.
