M modal.com >1B tokens/minute/GPU by combining query planner and inference engine 7 points charles_irl 1 month ago 2 comments
