Brand Logo
Back to workshops

How to run OpenAI’s open weight models locally on your computer

Timestamps:

[00:02] Introduce workshop topic and recording access
[01:32] Share screen for live demos
[02:12] Discuss new GPT-5 impressions
[03:16] Outline GPT-OSS session goals
[05:23] Recap OpenAI model history
[07:15] Explain reasoning model concept
[08:59] Highlight first open-source release in 5 years
[10:19] Compare 120B vs 20B model specs
[13:43] Review benchmark performance results
[18:24] Emphasize offline use and privacy benefits
[26:26] Demonstrate running models with Ollama
[33:09] Showcase LM Studio interface and features
[46:24] Suggest smaller models for lower-spec machines
[47:58] Present online GPT-OSS playground option
[50:25] Demonstrate speed using Groq platform
[60:42] Conclude with Q&A and future workshop ideas

Comments (2)

You need an active subscription to comment.

avatar

SCHUSTER JUERGEN CHRISTOPH

8/10/2025

Still don't get a grasp of how to fine tune a model. Wouldn't it be time Alvaro to finally fulfill your promise 😊

avatar

SCHUSTER JUERGEN CHRISTOPH

8/10/2025

How can a locally Open Source Model snitch to the government like shown in this benchmark: https://youtu.be/pnMgVIpBNf8?si=n_oZB_CkIvlTyvwV&t=764 It means they logging your input anyway and send it to their masters whenever they see fit.