Simon Willison • 11/23/2025

Olmo 3 is a fully open LLM

Read Original

The Allen Institute for AI (AI2) released Olmo 3, a series of fully open large language models. Unlike typical open-weight models, Olmo 3 provides complete transparency including training data (Dolma 3 corpus), code, and checkpoints. It features a 32B 'thinking' model that allows inspection of intermediate reasoning traces, plus 7B variants. The article details its training on 5.9 trillion tokens with enhanced code/math data and includes a hands-on test generating SVG code.

0 comments
#open source #AI Research #Training Data
Olmo 3 is a fully open LLM

Comments

No comments yet

Be the first to share your thoughts!

Browser Extension

Get instant access to AllDevBlogs from your browser

Top of the Week

1
Introducing GPT-5.1 for developers
Simon Willison • 6 votes
2
🧠 Build an Agent Chat that Remembers — Persisting Conversations with Microsoft Agent Framework
Bruno Capuano • 3 votes
3
Using A Hidden Submit Button To Ensure Unnamed Submissions
Ben Nadel • 3 votes
4
ServiceNow and Microsoft Copilot
Marius Sandbu • 2 votes
5
Cursor 2.0 and Composer Model
Codeaholicguy • 2 votes
6
Agentic AI and Security
Martin Fowler • 2 votes
7
Springs and Bounces in Native CSS
Josh Comeau • 2 votes
8
🚀 AG-UI + Agent Framework + .NET + Aspire: Web-Enabling Your Intelligent Agents (Blog + Demo + Code!)
Bruno Capuano • 1 votes
9
Fix “This video format is not supported” on YouTube TV
David Walsh • 1 votes
10
Tooltip Components Should Not Exist
TkDodo Dominik Dorfmeister • 1 votes