The idea of deploying large language models locally to save money and protect data privacy is great!

But diving into the world of models, the various parameters and model numbers can be overwhelming: 7B, 14B, 32B, 70B... The same model has so many parameters, which one should you choose?

And what level is my computer at? Which one can it run?

Don't panic! This article will help you clarify your thoughts and tell you in the simplest way how to choose hardware for deploying large models locally! Guarantee you won't be confused after reading it!

There is a Hardware Configuration and Model Size Reference Table at the bottom of this article.

Understanding Large Language Model Parameters: What Do 7B, 14B, and 32B Represent?

• Meaning of Parameters: The numbers 7B, 14B, 32B represent the number of parameters in a large language model (LLM), where "B" is an abbreviation for Billion. Parameters can be considered as the "weights" that the model learns during training, and they store the model's understanding of language, knowledge, and patterns.
• Parameter Count and Model Capability: Generally speaking, the more parameters a model has, the more complex the model is. Theoretically, it can learn and store richer information, thereby capturing more complex language patterns and performing more powerfully in understanding and generating text.
• Resource Consumption and Model Size: Models with more parameters also mean that they require more computing resources (GPU computing power), more memory (VRAM and system memory RAM), and more data for training and running.
• Small Models vs. Large Models:
  • Large Models (such as 32B, 65B or even larger): Can handle more complex tasks, generate more coherent and nuanced text, and may perform better in knowledge question answering, creative writing, etc. However, they have high hardware requirements and run relatively slowly.
  • Small Models (such as 7B, 13B): Consume fewer resources and run faster, making them more suitable for running on devices with limited resources or application scenarios that are sensitive to latency. Small models can also perform well on some simple tasks.
• Trade-offs in Selection: Choosing the size of the model requires a trade-off between the model's capabilities and hardware resources. More parameters are not necessarily "better"; the most suitable model needs to be selected according to the actual application scenario and hardware conditions.

What Kind of Hardware Do I Need to Run Local Models?

• Core Requirement: Video RAM (VRAM)

  • Importance of VRAM: When running large models, the model's parameters and intermediate calculation results need to be loaded into the video memory. Therefore, the size of the video memory is the most critical hardware indicator for running local large models. Insufficient video memory will cause the model to fail to load, or only very small models can be used, or even severely reduce the running speed.
  • The bigger, the better: Ideally, having a GPU with as much VRAM as possible is the best, so that you can run models with larger parameters and get better performance.

• Second most important: System Memory (RAM)

  • Role of RAM: System memory RAM is used to load the operating system, run programs, and serve as a supplement to video memory. When video memory is insufficient, system RAM can be used as "overflow" space, but the speed will be much slower (because RAM is much slower than VRAM), and the model running efficiency will be significantly reduced.
  • Sufficient RAM is also important: It is recommended to have at least 16GB or even 32GB or more of system RAM, especially when your GPU video memory is limited, more RAM can help alleviate video memory pressure.

• Processor (CPU)

  • Role of CPU: The CPU is mainly responsible for data pre-processing, model loading, and some model calculation tasks (especially in the case of CPU offloading). A CPU with good performance can improve the model loading speed and assist the GPU in calculations to a certain extent.
  • NPU (Neural Processing Unit): The NPU (Neural Processing Unit) equipped in some laptops is a hardware specially used to accelerate AI calculations. NPU can accelerate specific types of AI operations, including the reasoning process of some large models, thereby improving efficiency and reducing power consumption. If your laptop has an NPU, that would be a plus, but the GPU is still the core of running local large models. The support and effect of NPU depend on the specific model and software.

• Storage (Hard Disk/SSD)

  • Role of Storage: You need enough hard drive space to store model files. The files of large models are usually very large. For example, a quantized 7B model may also require 4-5GB of space, and larger models require tens or even hundreds of GB of space.
  • SSD is better than HDD: Using a solid-state drive (SSD) instead of a mechanical hard drive (HDD) can significantly speed up model loading.

Hardware Priority

1. Video RAM (VRAM) (Most Important)
2. System Memory (RAM) (Important)
3. GPU Performance (Computing Power) (Important)
4. CPU Performance (Auxiliary Role)
5. Storage Speed (SSD is better than HDD)

What if there is no Discrete GPU?

• Integrated Graphics and CPU Running: If you don't have a discrete GPU, you can still use integrated graphics (such as Intel Iris Xe) or rely entirely on the CPU to run the model. However, the performance will be greatly limited. It is recommended to focus on running 7B or even smaller, highly optimized models, and use technologies such as quantization to reduce resource requirements.
• Cloud Services: If you need to run large models but your local hardware is insufficient, you can consider using cloud GPU services such as Google Colab, AWS SageMaker, RunPod, etc.

How to Run Local Models?

For beginners, it is recommended to use some user-friendly tools that simplify the process of running local models:

• Ollama: Operated through the command line, but it is very simple to install and use, and focuses on running models quickly.
• LM Studio: The interface is simple and intuitive, supports model download, model management, and one-click running.

Hardware Configuration and Model Size Reference Table

Swipe left and right to see the full table

X86 Laptop
Integrated graphics laptop (such as Intel Iris Xe)	Shared system memory (8GB+ RAM)	8-bit, or even 4-bit quantization	≤ 7B (Extremely Quantized)	* Very basic local running experience, suitable for learning and light experience. * Limited performance, slow reasoning speed. * It is recommended to use 4-bit or lower precision quantization models to minimize video memory usage. * Suitable for running small models, such as TinyLlama, etc.
Entry-level gaming laptop/thin and light independent graphics laptop (such as RTX 3050/4050)	4-8 GB VRAM + 16GB+ RAM	4-bit - 8-bit quantization	7B - 13B (Quantized)	* 7B models can be run relatively smoothly, and some 13B models can also be run through quantization and optimization. * Suitable for experiencing some mainstream small and medium-sized models. * Note that VRAM is still limited, and running large models will be more difficult.
Mid-to-high-end gaming laptop/mobile workstation (such as RTX 3060/3070/4060)	8-16 GB VRAM + 16GB+ RAM	4-bit - 16-bit (flexible choice)	7B - 30B (Quantized)	* Can run 7B and 13B models more comfortably, and has the potential to try models around 30B (requires better quantization and optimization). * You can choose different quantization accuracy according to your needs to achieve a balance between performance and model quality. * Suitable for exploring more types of medium and large models.

---

ARM (Apple M Series)
Raspberry Pi 4/5	4-8 GB RAM	4-bit quantization (or lower)	≤ 7B (Extremely Quantized)	* Limited by memory and computing power, it is mainly used to run extremely small models or as an experimental platform. * Suitable for researching model quantization and optimization techniques.
Apple M1/M2/M3 (Unified Memory)	8GB - 64GB Unified Memory	4-bit - 16-bit (flexible choice)	7B - 30B+ (Quantized)	* The unified memory architecture makes memory utilization more efficient. Even M series Macs with 8GB of memory can run models of a certain size. * Higher memory versions (16GB+) can run larger models and can even try models above 30B. * Apple chips have advantages in energy efficiency.

---

Nvidia GPU Computer
Entry-level discrete graphics card (such as RTX 4060/4060Ti)	8-16 GB VRAM	4-bit - 16-bit (flexible choice)	7B - 30B (Quantized)	* Similar to the performance of mid-to-high-end gaming laptops, but desktop computers have better heat dissipation and can run stably for a long time. * High cost performance, suitable for entry-level local LLM players.
Mid-range discrete graphics card (such as RTX 4070/4070Ti/4080)	12-16 GB VRAM	4-bit - 16-bit (flexible choice)	7B - 30B+ (Quantized)	* Can run medium and large models more smoothly and has the potential to try models with larger parameters. * Suitable for users who have high requirements for local LLM experience.
High-end discrete graphics card (such as RTX 3090/4090, RTX 6000 Ada)	24-48 GB VRAM	8-bit - 32-bit (or even higher)	7B - 70B+ (Quantized/Native)	* Can run most open source LLMs, including large models (such as 65B, 70B). * You can try higher bit precision (such as 16-bit, 32-bit) to obtain the best model quality, or use quantization to run larger models. * Suitable for professional developers, researchers and heavy LLM users.
Server-level GPU (such as A100, H100, A800, H800)	40GB - 80GB+ VRAM	16-bit - 32-bit (native precision)	30B - 175B+ (Native/Quantized)	* Designed for AI computing, with ultra-large video memory and extremely strong computing power. * Can run ultra-large models and even perform model training and fine-tuning. * Suitable for enterprise-level applications, large-scale model deployment and research institutions.

Table Supplement Instructions

• Quantization: Refers to reducing the numerical precision of model parameters, for example, from 16-bit floating point (float16) to 8-bit integer (int8) or 4-bit integer (int4). Quantization can significantly reduce model size and video memory usage, and accelerate reasoning speed, but may slightly reduce model accuracy.
• Extreme Quantization: Refers to using very low bit precision quantization, such as 3-bit or 2-bit. It can further reduce resource requirements, but the decline in model quality may be more obvious.
• Native: Refers to the model running at its original precision, such as float16 or bfloat16. You can obtain the best model quality, but the resource requirements are the highest.
• Quantized Parameter Range: The "Recommended LLM Parameter Range (after quantization)" in the table refers to the model parameter range that the hardware can roughly run smoothly under the premise of reasonable quantization. The actual model size and performance that can be run also depend on the specific model architecture, quantization degree, software optimization and other factors. The parameter range given here is for reference only.
• Unified Memory: The characteristic of Apple Silicon chips is that the CPU and GPU share the same physical memory, and the data exchange efficiency is higher.