# [Deploy DeepSeek-R1 on Arm Servers with llama.cpp](https://learn.arm.com/learning-paths/servers-and-cloud-computing/deepseek-cpu/)

## In this learning path

- [Introduction](https://learn.arm.com/learning-paths/servers-and-cloud-computing/deepseek-cpu/)
- [Run a DeepSeek-R1 chatbot on Arm servers](https://learn.arm.com/learning-paths/servers-and-cloud-computing/deepseek-cpu/deepseek-chatbot/)
- [Access the chatbot using the OpenAI-compatible API](https://learn.arm.com/learning-paths/servers-and-cloud-computing/deepseek-cpu/deepseek-server/)
- [Next Steps](https://learn.arm.com/learning-paths/servers-and-cloud-computing/deepseek-cpu/_next-steps/)

## About this Learning Path

| Skill level:      | Introductory      |
|-------------------|-------------------|
| Reading time:     | 30 min            |
| Last updated:     | 31 Jul 2026       |

| Author:           | Tianyu Li, Arm    |
|-------------------|-------------------|
| Arm IP:           | [Neoverse](https://support.arm.com/?tab=compute-ip&Product%20Type=Infrastructure%20Processors) |
| Tags:             | [ML](https://learn.arm.com/tag/ml), [AWS](https://learn.arm.com/tag/aws), [Microsoft Azure](https://learn.arm.com/tag/microsoft-azure), [Google Cloud](https://learn.arm.com/tag/google-cloud), [Oracle](https://learn.arm.com/tag/oracle), [Linux](https://learn.arm.com/tag/linux), [LLM](https://learn.arm.com/tag/llm), [Generative AI](https://learn.arm.com/tag/generative-ai), [Python](https://learn.arm.com/tag/python) |

### Who is this for?

This Learning Path is for developers who want to run DeepSeek-R1 on Arm-based servers.

### What will you learn?

Upon completion of this Learning Path, you will be able to:

- Clone and build `llama.cpp` on your Arm-based server.
- Download a pre-quantized DeepSeek-R1 model from Hugging Face.
- Run the model on your Arm CPU and benchmark its performance.

### Prerequisites

Before starting, you will need the following:

- An [Arm-based instance](https://learn.arm.com/learning-paths/servers-and-cloud-computing/csp/) from a cloud provider or an on-premise Arm server. This Learning Path was tested on an AWS Graviton4 r8g.24xlarge instance.

### Summary

You’ll build `llama.cpp` on an Arm-based Ubuntu server, download a pre-quantized DeepSeek-R1 model from Hugging Face, and run it for CPU inference. You’ll launch a persistent `llama.cpp` server with an OpenAI-compatible API, deploy a chatbot with the DeepSeek-R1 671B LLM, and send requests from the same or another machine. You’ll then confirm the setup with API responses and basic benchmarks.

### Frequently asked questions

#### What do I need before running the 671B model?
Use an Arm server running Ubuntu 24.04 LTS with at least 64 cores, 512 GB of RAM, and 400 GB of disk space. The Learning Path was tested on an AWS Graviton4 `r8g.24xlarge` instance.

#### Which DeepSeek-R1 model should I download?
Download a pre-quantized DeepSeek-R1 model from Hugging Face. Quantization enables efficient CPU inference with `llama.cpp`.

#### How do I know the llama.cpp server binary is available?
Running `make` during the build step creates the server executable. If it’s missing, repeat the build step before starting the server.

#### How do I access the model repeatedly without restarting it?
Start the `llama.cpp` server and use its OpenAI-compatible API to submit multiple requests. You can also send requests from another machine to the host running the server.

#### Why do I install jq?
The examples use `jq` to work with JSON. `jq` helps format and parse responses returned by the OpenAI-compatible API calls.