The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →You can run a DeepSeek R1 model on a Windows PC or Mac with Ollama or LM Studio, but the practical choice is usually a smaller distilled model—not the full 671B model. On iPhone, LM Studio documents a way to connect to a model hosted on your computer; that is not on-device iPhone inference. DeepSeek’s Android assistant app listing likewise does not establish that R1 runs on the phone, and a supported Android on-device setup is not established here.
What “DeepSeek R1” means for a local setup
DeepSeek R1 is a model family, not a single download sized for every computer or phone. It includes the full R1 model and smaller distilled models. A distilled model is the sensible starting point for most personal computers: the full model’s download is hundreds of gigabytes, while smaller variants have substantially lower download sizes.
| # | Preview | Product | Price | |
|---|---|---|---|---|
| 1 |
|
GMKtec EVO-X2 AI Mini PC Ryzen Al Max+ 395 Superchip 128GB LPDDR5X 2TB SSD | $3,649.99 | Buy on Amazon |
Ollama’s catalog lists downloadable model files, not minimum RAM requirements or performance guarantees. The amount of memory needed while running a model also depends on factors such as quantization, context length, and other active workloads. A file fitting on disk does not prove that a device can run it well.
Choose a model size before downloading
The following download sizes are listed in Ollama’s model catalog, checked in 2026. They describe the downloads, not measured runtime memory, speed, or quality on a particular device.
#1 Best Overall
- EVOLUTION RYZEN AI MAX+ 395 MINI PC - GMKtec EVO-X2 is the next evolution in AI mini PC Ryzen Strix Halo series. Thanks to AMD Simultaneous Multithreading (SMT) the core-count is effectively doubled, to 32 threads. Ryzen AI Max+ 395 has 64 MB of L3 cache and can boost up to 5.1 GHz, depending on the workload. The Ryzen AI Max+ 395 is currently rated as the "most powerful x86 APU" on the market for AI computing.
- AI NPU with XDNA 2 ARCHITECTURE - Powered by 16 “Zen 5” CPU cores, 50+ peak AI TOPS XDNA 2 NPU and a truly massive integrated GPU driven by 40 AMD RDNA 3.5 CUs, the Ryzen AI MAX+ 395 is a transformative upgrade and delivers a significant performance boost over the competition. The Ryzen AI Max+ 395 excels in consumer AI workloads like the llama.cpp-powered application: LM Studio. Shaping up to be the must-have app for client LLM workloads, LM Studio allows users to locally run the latest language model without any technical knowledge required and unleash their creativity and productivity.
- AMD RADEON 8090S iGPU GAMING PC - The AMD Radeon RX 8060S offers all 40 CUs with up to 2.9 GHz graphics clock and uses the new RDNA 3.5 architecture. The powerful iGPU is positioned between an RTX 4060 and 4070 laptop GPU and therefore enables gaming in FHD at maximum details in most demanding games. The 8060S can also utilize the full 128GB pool, which is perfect for running LLMs such as Deepseek 70B Q8, which runs comfortably on this machine.
- EIGHT CHANNEL LPDDR5X - LPDDR5X is a new ground breaking memory small form factor installed on-board. With blazing speeds up to to 8000MT/s, it runs 1.5x faster than the DDR5 SODIMMs; 90% better performance over DDR5 SODIMMs in video conferencing and photo editing; 30% better performance in productivity apps; 12% better performance in digital content workloads.
- QUAD SCREEN 8K DISPLAY SUPPORT - EVO-X2 AI Mini PC support 4-screen 4K/8K output via HDMI 2.1 (8K@60Hz), DisplayPort 1.4 (4K@60Hz), and dual USB 4 40Gbps Transfer speed (supporting PD3.0/DP1.4/DATA). Ideal for gaming, video editing, and multitasking, it provides expansive and crisp multi-display support.
| Ollama model listing | Download size | Practical consideration |
|---|---|---|
| DeepSeek-R1-Distill-Qwen 1.5B | 1.1 GB | Smallest listed option here; the file size alone does not establish compatibility with a phone or a particular runtime. |
| DeepSeek-R1-Distill-Qwen 7B | 4.7 GB | A smaller distilled choice than the 14B and 32B variants. |
| DeepSeek-R1-0528 Qwen3 8B | 5.2 GB | An 8B listing; it is distinct from the Distill-Qwen 7B variant. |
| DeepSeek-R1-Distill-Qwen 14B | 9.0 GB | Requires more download and storage space than the smaller options. |
| DeepSeek-R1-Distill-Qwen 32B | 20 GB | Plan for substantial free storage and suitable system resources. |
| DeepSeek-R1-Distill-Llama 70B | 43 GB | A large download; do not infer that a computer can run it from storage capacity alone. |
| Full DeepSeek-R1 671B | 404 GB | Very large download and generally impractical for an ordinary personal-device setup. |
Pick a size that leaves room for the model file and other applications, then check how the chosen runtime behaves on your hardware. Do not treat a particular RAM amount as a guarantee of speed, usable context length, or successful operation. The catalog figures above are from Ollama’s 2026 listing, not a consumer-hardware benchmark.
Run DeepSeek R1 on Windows
Option 1: Ollama command line
- Install Ollama for Windows using its official installation route.
- Open a terminal, such as PowerShell or Windows Terminal.
- Run
ollama run deepseek-r1to use the catalog’s default DeepSeek R1 selection. - To request a particular listed size, use an explicit tag, for example
ollama run deepseek-r1:8b. Ollama also lists tags such asdeepseek-r1:1.5b. - Wait for the model download to finish, then enter a prompt in the running session.
The first run must retrieve the model files. The download size is not a promise about runtime memory or response speed.
Option 2: LM Studio desktop app
LM Studio documents DeepSeek R1 support on x64 and ARM64 Windows PCs. Install LM Studio, find a DeepSeek R1 variant in its model catalog, download a size appropriate for your machine, and load it in the app to chat. The exact model choice and runtime available can depend on the computer and selected model format; avoid assuming that every Windows PC can run every variant.
Run DeepSeek R1 on macOS
LM Studio’s download information lists Apple Silicon Macs and macOS 13 or later. Its documentation describes DeepSeek R1 support with llama.cpp and the MLX runtime on Apple Silicon. Ollama is another desktop option and uses the same commands shown above.
Free tools Windows power users keep installed
One-click scans. No signup required.
- Choose Ollama for a terminal-based setup or LM Studio for a graphical workflow.
- Install the selected app, then choose a distilled model size that fits your available storage and resources.
- For Ollama, run
ollama run deepseek-r1, or specify a tag such asollama run deepseek-r1:8b. - For LM Studio, download and load the chosen model in the app before starting a chat.
LM Studio documents that downloaded models can be used offline. Downloading is still required first; offline operation does not mean the model is already present on the Mac. Do not assume that every Mac can run the full 671B model or that a given memory configuration guarantees a particular speed.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Can you run DeepSeek R1 locally on Android?
DeepSeek’s official download page lists an Android assistant app, but that listing does not establish that the app downloads R1 model weights or performs inference entirely on the phone. Ollama’s 1.1 GB download listing for the 1.5B distilled model also does not establish that it works with a particular Android runtime or phone.
A verified Android on-device installation path and device requirements are not established here, so do not treat installing the official assistant app as proof of local inference. If the requirement is that prompts and model execution stay on the Android device, confirm that a current Android runtime explicitly supports the model and device before relying on it.
Use DeepSeek R1 from an iPhone through a computer
DeepSeek lists an official iOS assistant app, but an app listing is not evidence that R1 model weights run on the iPhone. LM Studio documents a different approach: run the model on a computer, link LM Studio through LM Link, then access it on the iPhone using the Locally app. In that arrangement, the computer hosts the model and the iPhone connects to it; it is computer-hosted access, not inference performed entirely on the iPhone.
Recommended Free Tools
- Install LM Studio on a supported Windows PC or Apple Silicon Mac, and download and load a DeepSeek R1 model there.
- Set up LM Link in LM Studio so the computer can be linked for remote access.
- Use the Locally app on the iPhone to connect to the linked computer and access its hosted model.
This route depends on the computer being available to host the model. It should not be described as running R1 locally on the iPhone itself.
Advanced desktop options and model licensing
DeepSeek’s repository gives local-serving examples using vLLM and SGLang for a 32B distilled model. It also cautions that Transformers was not directly supported in the repository text, so do not assume that a generic Transformers setup is an officially documented route. These serving tools are more involved than the Ollama or LM Studio paths and are aimed at users comfortable configuring model-serving software.
DeepSeek’s January 20, 2025 release announcement describes R1 as a “Fully open-source model & technical report.” The repository states an MIT license, but some distilled models are based on other models with their own underlying licenses. Check the applicable repository and base-model terms for the specific variant you download rather than extending one licensing statement to every model in the family.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
