Intelligence Cost-Performance

Microsoft added average token usage to its model release card; the model scored 71.6 on SWE-Bench Verified while using about one-third of Claude Haiku 4.5’s tokens.

Why it matters: HKR-H/K/R all pass: the score-per-token angle is clickable, with concrete 71.6 and one-third-token claims. The article is thin on full test setup and pricing, so it lands at 78.