Mac Studio
A high-performance desktop computer from Apple equipped with M-series Ultra chips and a large unified memory architecture, suitable for high-load tasks such as running large AI models locally.
Core Facts
The new Mac Studio uses a unified memory design, with the 128 GB memory version starting at RMB 19,999 for the domestic Chinese market and the 512 GB version starting at RMB 46,999.
A Mac Studio M4 Ultra equipped with 192 GB of unified memory can theoretically run complete 70B or even larger parameter models.
苹果芯片采用统一内存设计，CPU、GPU与神经网络引擎共享同一物理内存池，内存带宽可达300GB/s以上
Timeline (last 90 days)
When two 256GB Mac Studios are connected via Thunderbolt or Ethernet, the inter-machine bandwidth (on the order of tens of Gbps) is far lower than a single machine's internal unified memory bandwidth (on the order of hundreds of GB/s).
Deploying Meta's model requires two top-spec Mac Studios, each with 512GB of memory and 1TB of storage.
Meta released a large open-source model and said it can run on two Mac Studio desktop machines.
Early versions of Hishob ran on a 6GB mini PC using a small 3B model; the hardware was later upgraded to a Mac Studio (M3 Ultra, 96GB unified memory) running the Qwen3.5-122B-A10B model locally.
The blogger koustubh runs the 4-bit GGUF quantized version (UD-IQ4_NL) of Qwen3.5-122B-A10B on a Mac Studio, actually occupying about 62GB of memory.
The Mac Studio is configured with an M3 Ultra chip and 96GB of unified memory, leaving more than 30GB to simultaneously run Immich, Nextcloud and a Kubernetes cluster.
gharasathi currently runs Qwen3.5-122B on the Mac Studio as the household financial analyst.
Models at the 122B level usually need quantization to run on the Mac Studio's unified memory, but they perform significantly better than 3B-level models at complex reasoning and following constraints.
gharasathi is an AI system that runs entirely inside the home, integrating money, photos and documents, with inference running on a single Mac Studio and data never sent to external clouds.
Apple updated its MLX toolkit to ensure MuseSpark Open runs smoothly on its own hardware, and Zuckerberg posted a video of Muse running on a Mac Studio and thanked Apple's John Ternus.
40 more timeline events
All Facts (20)
The new Mac Studio uses a unified memory design, with the 128 GB memory version starting at RMB 19,999 for the domestic Chinese market and the 512 GB version starting at RMB 46,999.
70%VerifiedA Mac Studio M4 Ultra equipped with 192 GB of unified memory can theoretically run complete 70B or even larger parameter models.
70%Verified苹果芯片采用统一内存设计，CPU、GPU与神经网络引擎共享同一物理内存池，内存带宽可达300GB/s以上
65%VerifiedAfter 4-bit quantization, Llama 3.3 70B has a model size of approximately 40 GB and can run on a Mac Studio with 64 GB of memory or a single professional GPU with 48 GB of VRAM.
65%VerifiedThe Mac Studio M5 Ultra starts at $5,499, with up to 512GB of memory and memory bandwidth of 1.2TB per second, with the 512GB version expected to ship in late October.
80%VerifiedApple introduced a new generation Mac mini equipped with the M6 chip.
80%VerifiedThe Mac mini M6 starts at $899, with up to 32GB of unified memory, going on sale September 22, 2026.
75%UnverifiedUpCtl在实际部署中采用混合云架构，Agent部署在本地Mac Studio上，Service层和Web端部署在公网云主机上
90%Partially VerifiedMuse Spark Open can run on two Mac Studios equipped with M5 Ultra chips, together with the latest MLX framework.
65%UnverifiedMuse Spark Open is the first model of its kind that can run locally on two Mac Studios with Apple M5 Ultra chips with the help of the MLX framework.
60%UnverifiedThe M5 Ultra version of the Mac Studio is equipped with up to 80-core GPU and up to 512GB of unified memory
60%UnverifiedThe Mac Studio is equipped with six Thunderbolt 5 ports, HDMI 2.1, a 10G Ethernet port, and retains full-size USB-A ports and an SDXC card reader.
60%UnverifiedIn AmorphousDiskMark tests, the M5 Ultra achieves sequential read speeds of 14,888 MB/s and write speeds approaching 20,000 MB/s, while the M3 Ultra reads at approximately 6,500 MB/s and writes at 2,700 MB/s.
60%UnverifiedAccording to a report cited in the video, OpenAI purchased tens of thousands of Mac mini and Mac Studio units for reinforcement learning, allowing AI to practice computer operations on real machines.
60%UnverifiedThe Mac Studio is equipped with M-series chips, such as the M2 Max, M2 Ultra, and subsequent updated versions.
60%UnverifiedWhen two 256GB Mac Studios are connected via Thunderbolt or Ethernet, the inter-machine bandwidth (on the order of tens of Gbps) is far lower than a single machine's internal unified memory bandwidth (on the order of hundreds of GB/s).
50%UnverifiedDeploying Meta's model requires two top-spec Mac Studios, each with 512GB of memory and 1TB of storage.
50%UnverifiedEarly versions of Hishob ran on a 6GB mini PC using a small 3B model; the hardware was later upgraded to a Mac Studio (M3 Ultra, 96GB unified memory) running the Qwen3.5-122B-A10B model locally.
50%UnverifiedThe Mac Studio is configured with an M3 Ultra chip and 96GB of unified memory, leaving more than 30GB to simultaneously run Immich, Nextcloud and a Kubernetes cluster.
50%UnverifiedThe blogger koustubh runs the 4-bit GGUF quantized version (UD-IQ4_NL) of Qwen3.5-122B-A10B on a Mac Studio, actually occupying about 62GB of memory.
50%