Mythic

One-liner: Fast low-level PDF engine: render, extract, manipulate (+ PyMuPDF4LLM)

One-liner: The classic open-source OCR engine powering many doc pipelines

One-liner: Precision PDF layout analysis: char-level geometry and table extraction

One-liner: Opinionated production model serving/training with TensorRT-LLM engines

One-liner: Serverless GPU compute with per-second billing and Python decorators

One-liner: Document ETL suite: partitioning, chunking, 20+ source connectors to vector DBs

One-liner: Marketplace of thousands of community models, pay-per-run

One-liner: Fastest-inference open-model serving with Tinker RL fine-tuning