xAI reported a capability gain for Grok 4.6: an Artificial Analysis Intelligence Index score of 61 at high reasoning, versus 56 for Grok 4.5 at the same named setting. Its launch article also listed starting API rates of $2 per million input tokenThe chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.Full definition → and $6 per million output tokens.
Those rates alone cannot settle whether the improvement is worth its cost. Artificial Analysis says its cost per task is calculated from token prices across input, cache, reasoning and answer tokens, so usage matters as well as price. A model can become more capable while spending more on each attempt, so the meaningful comparison pairs task results with the cost of obtaining them.
Terms in this piece · Glossary
token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
