Ollama Blog
AI news source. Articles are auto-selected and adapted by Hamidun News editors.
Latest publications

Ollama speeds up Gemma 4 on Mac by nearly 90% in Aider coding-agent test
Ollama 0.31 speeds up Gemma 4 on Apple Silicon by nearly 90% in the Aider polyglot test, where a coding agent solves real tasks. The model's answers do not change.

Ollama launch runs Claude Code and Codex on local models without config files
The new ollama launch command configures Claude Code, OpenCode, and Codex for a local or cloud model automatically. Running locally requires about 23 GB of VRAM.

Ollama Adds Text-to-Speech on Mac: Phonon Runs 35% Faster Than Gradium's Native Runtime
Ollama can now read text aloud, with Phonon from Gradium as its first speech model. On a MacBook Pro with M5 Max, it generates speech 35% faster than Gradium's native ptts runtime and works through the OpenAI API.

Ollama on Mac Switches to MLX: 1851 Tokens/s Prompt Processing in Version 0.19
Ollama's preview release on Mac now runs on MLX, Apple's framework. Version 0.19 claims 1851 tokens/s for prompt processing and 134 tokens/s for response generation.

Ollama Speeds Up NVIDIA Model Runs by Up to 20% and Turns On Vulkan for AMD and Intel
Ollama improved its GGUF handling: on NVIDIA GPUs, models run up to 20% faster, and Vulkan now connects AMD and Intel cards to local AI models by default.

Stanford's OpenJarvis Brings a Personal AI Agent to Your Own Hardware via Ollama
Stanford's OpenJarvis framework builds a personal AI agent that runs on your computer, not in the cloud. It now installs through Ollama with a single command.

Reflection Releases Beam, a 501B-Parameter Open Model, Free on Ollama
Reflection's first open-weights model targets code and agents: 501B total parameters, 23B active per token, and a 1M-token context. Ollama is offering it for free for now.