Does Deepseek harness handle screenshot and the vision model in general?

When i tried giving the harness a screenshot it just said that i should pick a model with vision, but i was for sure on the vision model. so it let me to think that it does not yet handle vision in the harness?