How do you install and run DeepSeek and other LLMs locally on your own computer?
Running LLMs locally via tools like Ollama and LM Studio gives professionals full data privacy, no API costs, and offline capability for sensitive financial work.
- Local LLM deployment prevents confidential client data from being transmitted to external servers - Tools like Ollama and LM Studio enable non-engineers to run models like DeepSeek locally on standard hardware - Local inference eliminates API costs and enables offline operation for sensitive professional work - Performance depends on hardware, particularly GPU VRAM, but modern laptops can run smaller models effectively - Financial and legal professionals should evaluate local LLM deployment as a data governance and compliance tool
Artificial intelligence has made significant strides in the development of large language models. While many of these models are hosted on cloud-based platforms, there is growing interest in running them locally on personal computers. Running an LLM locally provides greater privacy, reduced latency, and the ability to operate without an internet connection or API costs.
Why Run a Model Locally
Cloud-hosted LLMs process your queries on external servers, meaning every question you ask is transmitted, logged, and potentially used for model improvement. For professionals handling confidential client information, proprietary analysis, or sensitive financial data, this creates a genuine data governance concern. Local models process everything on your own hardware. Nothing leaves your machine.
The cost argument is also compelling. At scale, API costs for frequent or complex queries accumulate quickly. A locally hosted model has zero marginal cost per query once installed.
What You Need
Modern consumer hardware is capable of running small-to-medium sized models reasonably well. A machine with at least 16GB of RAM can run quantized versions of models like DeepSeek R1, LLaMA, or Mistral at acceptable speed. Machines with 32GB or more of RAM, or with a dedicated GPU, will perform significantly better. The model size and quantization level determine the hardware requirements.
Step-by-Step Installation with Ollama
The simplest way to run models locally is through Ollama, an open-source tool that manages model downloads and inference. Download Ollama from ollama.com and run the installer for your operating system. Once installed, open a terminal and run the command: ollama run deepseek-r1. Ollama will download the model automatically and launch an interactive chat interface in your terminal. For a web-based interface, install Open WebUI, which provides a ChatGPT-style browser interface that connects to your locally running Ollama instance.
Practical Applications
Locally running LLMs are well suited for analyzing confidential documents, drafting internal memos, summarizing financial filings, and building custom applications that require AI inference without external API dependencies. For investment banking and advisory professionals, the ability to analyze sensitive deal documents without transmitting them to external servers represents a meaningful security and compliance advantage.
Running large language models locally has moved from a developer curiosity to a practical imperative for professionals handling confidential client data. Cloud-hosted models transmit queries to external servers where data may be logged or used for training, creating real exposure for investment bankers, lawyers, and executives working with material non-public information. Tools like Ollama and LM Studio now make local deployment accessible to non-engineers on standard laptops, enabling private AI inference on documents, analysis, and communications. For organizations concerned about data governance, local LLM deployment provides a compliance-compatible alternative to cloud-based AI services.
When to speak with Chatsworth
You may benefit from an advisory conversation if your board is evaluating timing, valuation expectations, buyer universe quality, or diligence readiness. Chatsworth provides senior-led perspective on process design and execution risk independently of whether a mandate results.
Speak with the team →