CompressionBench evaluates whether coding agents can autonomously compress LLMs

November 5, 2026 · 3m

Bit puts a local LLM, vision, and image stack on Android — with a cloud escape hatch

Bit is an Android assistant that runs GGUF text, vision, image generation, and local RAG on-device via llama.kt, and can fall back to OpenAI-compatible APIs when a model won't fit.