Calibrated on 193 published benchmarks

Vision models you can run

Models that read images, screenshots and documents along with text.