Spark, a lightweight real-time coding model powered by Cerebras hardware and optimized for ultra-low latency performance.
This desktop app for hosting and running LLMs locally is rough in a few spots, but still useful right out of the box.