Someone needs to tell me why Grok 4 with Vision, a very powerful model, is at the bottom? In fact, the test depends on vision - yet all models perform poorly w

Someone needs to tell me why Grok 4 with Vision, a very powerful model, is at the bottom? In fact, the test depends on vision - yet all models perform poorly with that capability?