Create

Sign in to ReadmeX

or

How we handle your data: Privacy Policy

ReadmeX

Your AI reading companion

What would you like to catch up on?

Pick a briefing, or ask your own question.

Your community and people briefings stay personal to you.

Mac Studio

AI overview

Sign in and the AI will write an overview from our coverage.

Headlines · 6

  1. Apple and Meta team up to run Muse Spark Open locally on Mac Studio

    According to an analysis published on Huxiu, Meta's open-weight model Muse Spark Open will be able to run locally on clusters of M5 Ultra-equipped Mac Studio machines, with Apple providing on-device inference through its MLX framework and multi-machine coordination. The article says the model has roughly 816 billion total parameters and about 42 billion active parameters, and that Apple's machine learning team found it needs more than 1.6TB of memory at native precision and still about 493GB with 4-bit quantization, requiring several Mac Studios in parallel. It places the move alongside Apple's earlier integration of ChatGPT into Siri and of Google Gemini into Apple Intelligence, arguing Apple wants more models to run well on its hardware while Meta trades open weights for developer adoption and enterprise deployment.

    虎嗅 AI · 🔥 11
  2. Apple's MLX adds RDMA-distributed inference across Mac Studio clusters

    Apple's machine learning research team published benchmarks showing how the latest MLX release, combined with the low-latency RDMA over Thunderbolt 5 driver added in macOS 26.2, can run frontier-scale open-weight models across multiple Mac Studio computers. On four M5 Ultra Mac Studios, the 816B-parameter Muse Spark Open mixture-of-experts model (42B active, NVFP4) generates over 36 tokens per second at an 8,192-token context, while a single 512GB Mac Studio runs Qwen 3.8 27B and Mistral Medium 3.5 128B. Apple reports the M5 Ultra improves time to first token by up to 4.6x and decode speed by 1.6x versus M3 Ultra, and says pipeline parallelization lifted prefill by up to about 1.8x at 65,536-token contexts.

    Apple Machine Learning Research · 🔥 3
  3. Four Tesla V100 GPUs beat two DGX Spark units in inference test

    The YouTube channel Jun Yamog posted an Oct 12 test running four 2017-era Nvidia Tesla V100 data-center GPUs on an NVLink carrier board with a 4-bit quantized Qwen3.8-Flash-Next model. The cluster was reported 40% faster than two DGX Spark units and 230% faster than an M5 Ultra Mac Studio on long-prompt processing, and 40% and 163% faster respectively on short prompts. Short-prompt token generation reached 100.6 tok/s, about 121% faster than the two DGX Sparks, while the M5 Ultra measured 44.5 tok/s. The tester said the large gap versus Apple Silicon may partly reflect weaker software optimization.

    IT之家 AI · 🔥 0
  4. GPUs in the wine cellar: the homelab boom in local AI compute

    A local-AI wave is growing as small developers and hardware hobbyists run high-end GPU machines in basements, attics and home offices for cost, privacy and convenience, cnBeta reports. Former Microsoft employee and repeat founder Gary Fleck built a machine with discounted parts for under $6,000 to train custom models on terabytes of EEG data, saying it beats paying monthly AWS or Azure bills — though cooling his basement required $10,000 to $20,000 in electrical work and HVAC. Local agent platforms such as OpenClaw are helping drive the trend, Mac mini and Mac Studio are popular with agent developers, and Dell says the home-AI hobbyist market is booming. Crazy Egg CEO Hiten Shah runs 18 devices including 14 Mac minis and Mac Studios, with his home office often hitting 85°F (29.4°C), while one YouTube creator runs Qwen-3.8-27B locally on five AMD GPUs in his garage. Some enthusiasts say they want to keep compute in their own hands in case governments restrict AI access.

    cnBeta · 🔥 0
  5. Hands on with local AI: Hermes Agent and Qwen on a Mac Studio

    A Verge reviewer installed the open-source Hermes Agent desktop app on an M5 Ultra Mac Studio with 256GB of unified memory and chose the roughly 105GB, 125-billion-parameter Qwen 3.8 Flash Next model to run locally. He used it to build a daily briefing, sort a Steam library of more than 400 games by genre, analyze financial records and make a laptop spec-comparison spreadsheet, and is now trying to script partial automation of laptop benchmark testing. The author says local models are no magic assistant — tasks still break — but the privacy of keeping data on the machine makes him more willing to trust it with sensitive work.

    The Verge AI · 🔥 0
  6. M5 Max Mac Studio drops $200 as Apple readies 512GB M5 Ultra for local AI

    TechRadar reports that the Mac Studio with M5 Max (18-core CPU, 32-core GPU, 16-core Neural Engine), 36GB of memory and 512GB of storage is down to $2,299 from $2,499 at B&H Photo, an 8% discount; the 1TB configuration is $2,579, down from $2,799. The article says Apple's 512GB M5 Ultra Mac Studio, aimed at running large AI models locally, is due to ship in late October and may appear at Apple's event next week, with the author speculating it could start well above $10,000. Citing a roundup of M5 Ultra reviews, it says the Ultra's real strength is local AI, while its multi-core lead over the M5 Max is only about 26% and single-core performance is roughly the same.

    TechRadar · 🔥 0

Experience and discussion from the community

Share my Mac Studio experienceAsk about Mac Studio

Nobody has shared their experience with Mac Studio yet.