v0.16.0
A new reasoning giant and SOTA coder just dropped for local AI.
Deep Dive
Ollama's v0.16.0 release introduces two major new models. The standout is GLM-5 from Z.ai, a 744B total parameter model (40B active) built for complex reasoning and long-horizon agentic tasks. It also adds MiniMax-M2.5, a new state-of-the-art model for coding and productivity. The update includes a new `ollama launch` command for apps and improvements to the MLX runner for Apple Silicon support.
Why It Matters
This brings frontier-scale reasoning and top-tier coding capabilities directly to developers' local machines for free.