Remarkable how few tokens it needed to get a much better score than other reason... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		causal 9 months ago \| parent \| context \| favorite \| on: Gemini 2.5 Remarkable how few tokens it needed to get a much better score than other reasoning models. Any chance of contamination?

85392_school 9 months ago [–]

It makes me wonder how the token counting was implemented and if it missed the (not sent in API) reasoning.

freediver 9 months ago | [–]

Vaild concern, most likely thinking tokens were not counted due to API reporting changes.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact