How to run OpenAI’s open weight models locally on your computer
Run open-weight reasoning models on your own computer with Ollama and LM Studio while balancing hardware limits, privacy, model size, and hosted testing options.
- Beginner
- 10 lessons
- 1 hr 2 min
- Taught by Dr. Alvaro Cintas
What you'll learn
- Pick the Right Model Choose between GPT-OSS, hosted OpenAI models, and other open models based on privacy, offline access, cost, customization, and model quality.
- Evaluate Hardware Fit Assess whether GPT-OSS 20B or 120B fits your machine by comparing memory, parameter size, architecture, benchmarks, and practical performance needs.
- Run Models with Ollama Install Ollama and run GPT-OSS locally from the terminal, with smaller model alternatives ready when your computer cannot handle larger models.
- Use LM Studio Locally Use LM Studio to download local models, chat through a visual interface, and query local documents with retrieval-augmented generation.
- Protect Private Data Decide when fine-tuning, retrieval, or local execution is appropriate for sensitive legal, healthcare, finance, research, coding, or internal business workflows.
- Experiment Safely Test hosted demos safely for quick experimentation while keeping sensitive prompts and files reserved for local workflows.
Lessons 10 lessons · 1 hr 2 min
Every lesson comes with Pro, $29 a month.
- Lesson 1: Workshop Introduction and Setup 3 min
- Lesson 2: GPT-5 Reactions and Session Outline 4 min
- Lesson 3: OpenAI Model Timeline and Reasoning Models 6 min
- Lesson 4: GPT-OSS Model Specs and Benchmarks 8 min
- Lesson 5: Why Open Source Matters 8 min
- Lesson 6: Fine-Tuning and Privacy Use Cases 6 min
- Lesson 7: Running GPT-OSS Locally with Ollama 7 min
- Lesson 8: Using LM Studio and File Uploads 7 min
- Lesson 9: Online Demo Options and Model Selection 8 min
- Lesson 10: Q&A and Closing Remarks 3 min
Your teacher
Dr. Alvaro Cintas
PhD AI Professor