Ollama Updates MLX Runtime and Supports New Models

Ollama's MLX runtime now runs model architectures on Apple Silicon devices by default, and the platform has added support for several new models.

By Precis Daily NewsroomOctober 7, 20261 min read117 words

Source:Ollama (GitHub releases) ↗

Illustration for: Ollama Updates MLX Runtime and Supports New Models — Illustration

Key points

Ollama's MLX now supports the models qwen3.8, gemma4, qwen3.6, and qwen3.5.
According to Ollama (GitHub releases), the company's MLX runtime now runs model architectures on Apple Silicon devices by default.
The addition of Apple Silicon support expands Ollama's user base and allows developers to integrate the platform's machine learning capabilities into their projects more easily.

What happened

According to Ollama (GitHub releases), the company's MLX runtime now runs model architectures on Apple Silicon devices by default. This update enables developers to deploy Ollama's machine learning models on Apple devices without modification.

Why it matters

The addition of Apple Silicon support expands Ollama's user base and allows developers to integrate the platform's machine learning capabilities into their projects more easily.

Key details

Ollama's MLX now supports the models qwen3.8, gemma4, qwen3.6, and qwen3.5. Additionally, decision models Nimble, tev1, clef, and clef-flash are available on the platform. The company has also added support for the embedding model, embeddinggemma-2. Ollama is continuing to test and enable additional models on its MLX runtime.