MLX

I'm Obsessed With Local AI. Here's Why

I run this episode solo. I explain local AI in plain terms: the model runs on hardware I control, and a cloud model runs somewhere else. I map the four pieces of the local AI landscape — the model, the warehouse, the software, and the workflow — and I define the words that beginners meet first:…

Fable 5.1 🧠, World Labs Atlas 🌍, Cognition $47B 💰

Atlas is a world generation model pretrained from scratch to natively operate on text, images, video, and 3D. It combines all inputs into a shared spatial context and uses that context to generate what comes next.

Qwen3.8-Max open weights: 4.89 TB, 224 files, one license catch

Alibaba finally opened a Max-class Qwen — and the 2.4T checkpoint downloads at 4.89 TB, three times Kimi K3 for a smaller model, because Qwen shipped BF16 where Moonshot shipped pre-quantized 4-bit.

Muse Glimmer from Meta Superintelligence Labs is now available

Meta's Muse Glimmer, the first open model released by Meta Superintelligence Labs, is now available. Muse Glimmer is a 30B multimodal model released under the Apache 2.0 license, designed for local coding agents, and accelerated by Ollama's MLX engine with new native DFlash and image input support.

Kimi K3 open weights: the self-hosting reality check

Moonshot released Kimi K3's weights on July 27, 2026. What that actually gets you at 2.8 trillion parameters — the real download size, the hardware floor, the bespoke license, and how to choose between self-hosting, the official API, and an aggregator.

From CUDA to MLX: How K-Search Brings Decades of Kernel Expertise to Apple Silicon

Figure 1: CUDA-to-MLX optimization translation map. CUDA optimization knowledge can be translated into architecture-native MLX strategies rather than copied instruction-for-instruction. We face a new epoch in computing.

Generating Test Data for Pinecone

A repeatable workflow for building large, realistic vector test datasets: CC News to Parquet to local embeddings to Pinecone bulk import.

Using Local Coding Agents

Using Open-Weight Models in Local Coding Harnesses as an Alternative to Claude Code and Codex Subscriptions

Ollama's highest performance on Apple Silicon yet with MLX

Ollama's MLX engine has been updated to deliver its highest performance on Apple Silicon yet. Models output higher quality responses, respond faster, and use less memory.

DiffusionGemma: 4x faster text generation

An overview of DiffusionGemma, an exceptionally fast text generation model with up to 4x faster speeds.

Improving LM Studio's MLX Engine for Agentic Workflows

mlx-engine v1.8.5 dramatically improves performance for repeated, long-context agentic workflows by checkpointing your KV cache.

Improved performance and model support with GGUF

Ollama 0.30 is now available with improved performance and GGUF model compatibility through llama.cpp. This augments Ollama's MLX engine on Apple silicon, bringing support to more models on a wider range of hardware.

Am I a Bad Friend?

I analysed 20 years of my chats and turned 1.2M messages into a structured vault of my life - to win friends and influence people. Instead, I learnt things about my emotional bandwidth, endearment cycles, and friendship half-lives.

Ollama is now powered by MLX on Apple Silicon in preview

Today, we're previewing the fastest way to run Ollama on Apple silicon, powered by MLX, Apple's machine learning framework.

Use your LM Studio Models in Claude Code

Run Claude Code with any local model using LM Studio's Anthropic-compatible API

Introducing LM Studio 0.4.0

Server deployment, parallel requests with continuous batching, new REST API endpoint, and refreshed application UI

LM Studio 0.3.38

LM Studio 0.3.32

GLM 4.5 tool calling, olmOCR-2, improved image input handling in `/v1/responses`, Flash Attention defaults for Vulkan/Metal, and bug fixes.