Jev beats Gemini 2.5 Flash Lite on classifier eval quality and speed
Malte Ubl ran TypeSafe AI's Jev against an existing classifier eval, where it saturated the eval on quality and was 6x faster than Gemini 2.5 Flash Lite.
The author asked Jev to assign excerpts from TypeSafe and OpenRouter press releases and doc pages to one of 13 categories, with an inspectable ChainForge flow shared.
To make this chart, we asked Jev which of 13 categories best describes each excerpt, taken from typesafe and OpenRouter's press releases and doc pages. "System One model" was one, "Decision model" was another. The flow is available to inspect right here: https://chainforge.ai/play/?f=161gf9kjcdygh…