OpenAI’s length adjustment moves each task’s score by 1.47 points for every 500 characters above or below 2,000 on HealthBench Professional, and by 2.99 points on HealthBench. Claude Opus 5.5 (max with fallback) has the highest raw HealthBench Professional score (71.4%) but averages ~5,900 characters, dropping it to sixth (59.9%). On HealthBench, Gemini 4 Argon (high) loses the most (11 points), falling from sixth on raw score (60.0%) to 13th (49.1%), with answers averaging ~3,800 characters
OpenAI’s length adjustment moves each task’s score by 1.47 points for every 500 characters above or below 2,000 on HealthBench Professional…
OpenAI’s length adjustment moves each task’s score by 1.47 points for every 500 characters above or below 2,000 on HealthBench Professional, and by 2.99 points on HealthBench. Claude Opus 5.5 (max with fallback) has the highest raw HealthBench Professional score (71.4%) but averages ~5,900 characters, dropping it to sixth (59.9%).
Artificial Analysis · X
Content notes
Original publication and rights belong to the source.