Open-source AI-SQL engine Quail hits 1B+ input tokens/min on a single H100

sh_reya · x · 2026-09-27

shreya unveils Quail, a new open-source AI-SQL engine built with Modal. By planning SQL queries and LLM inference together, it reaches over 1 billion input tokens per minute on a single H100 for one query. The team argues AI-powered data operators create a genuinely new class of inference workload; commenters note domain-specific harnesses, now a meme, can be incredible when done right.

Related event: UC Berkeley and Modal Open-Source Quail, an AI-SQL Inference Engine Hitting 1B+ Tokens/min on One H100(24 posts)→

Original post →

More from coding & agent

coding & agent channel →