Testing Jev multimodal: 408 calls on 48 photos
The author ran 408 Jev calls on 48 openly licensed photos with verifiable labels, spending under four cents to check how well image requests return probabilities, labels, levels or boxes and where they fail.
TypeSafe Raised $870M and made Jev multimodal model I ran it with 408 Calls on 48 Real Photos You can put a photo in a request and get back a probability, a label, a level, or a box. I sent 48 openly licensed photos through both Jev versions, twice each, and checked every answer against labels I could verify. The 384 calls, plus 24 more to test question wording, cost less than four cents. This is what typed answers look like on images, and where they break. https://aiengineeringfromscratch.com/blog/typesafe-870m-can-jev-see
This post quotes another post: x.com/typesafeai/status/2110536685805355064