Shipped some small quality of life improvements to kontext today. You can now: – change the safety_tolerance (though image inputs are still limited to 2)
– pick an output format (jpg or png) Also, we have two multi-image models now.

→ View original post on X — @fofrai,