Unchanged prices per 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 → can still produce a higher bill per task. In its August 5, 2026 evaluation, Artificial Analysis reported that Meta's Muse Spark 1.2 cost an average of $0.40 per Intelligence Index task, up from $0.29 for Muse Spark 1.1, while token prices stayed the same.
Artificial Analysis attributed the increase to usage: about 53% more input tokens and 36% more output tokens per task than 1.1, concentrated in GDPval-AA v2, an agentic knowledge-work evalA repeatable test for AI quality — a set of tasks plus scoring — used the way software teams use test suites, because model output is too variable to judge by eyeballing.Full definition →. On that evaluation 1.2 also scored higher, so the extra spending came with a measured gain. These are benchmarkA standard public test set for comparing AI models — the shared scoreboards behind every "model X beats model Y" claim.Full definition → costs for one test suite, not a forecast for any application's workload, and Meta gave Artificial Analysis pre-release access.
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.
eval — A repeatable test for AI quality — a set of tasks plus scoring — used the way software teams use test suites, because model output is too variable to judge by eyeballing.
benchmark — A standard public test set for comparing AI models — the shared scoreboards behind every "model X beats model Y" claim.
context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
