Osmosis

Companies struggle to post-train and fine-tune language models for specific tasks with better performance, cost, and latency than foundation models.

A CLI-based post-training platform that orchestrates compute, builds reward models, and runs observable reinforcement learning fine-tuning for language models.

Fast-growing AI companies and their developers building task- or domain-specific language models.