A person has spent years collecting everything worth reading on open models, and now they have published the list as a gift to anyone who wants to understand the field.
The collection covers economics, safety, adoption, and geopolitics, with papers dating back to 2023. It is a comprehensive overview of where open models stand today, and the compiler is asking readers to add their own recommendations in the comments.
Foundation Papers
Bill Gurley’s “From Open Source Software to Open Source Strategy” explains how open-source software has been used by businesses and what that means for AI. Mark Zuckerberg’s comments on Llama 3’s release, titled “Open Source AI is the Path Forward,” are among the clearest statements on the topic.
Irene Solaiman’s “The Gradient of Generative AI Release: Methods and Considerations” argues that open models exist on a gradient rather than a binary. Christian Catalini’s “Some Simple Economics of Open versus Closed AI” draws on IP history to explain how open models can capture value by serving as a complementary tool to the existing economy.
Nathan Lambert’s “Open models in perpetual catch-up” makes the case that open models will always lag behind closed ones in performance. His “Open and closed models are on different exponentials” explains how adoption differs between the two. Florian Brand’s “The Myth of unsafe Open Source AI” counters the idea that open models are uniquely dangerous, arguing that closed models regularly bypass safety guardrails.
The ATOM Project
The ATOM Project, led by Nathan Lambert, argues that the U.S. needs to invest in open models for fundamental R&D to keep pace with China. Lambert’s “Why I build open language models” explains why open models spur research innovation. Kevin Xu’s “Chinese Open Source: A Definitive History” and “China’s Structural Advantage in Open Source AI” lay out the history and the structural forces driving China’s position.
Adoption and Data
The Interconnects Adoption Dashboard tracks downloads, derivatives, and research adoption by region. The Interconnects Artifacts Hub lists the most important models in the ecosystem. The ATOM Report compares U.S. and Chinese model adoption.
| Paper | Author | Date |
|---|---|---|
| Gurley’s strategy paper | Bill Gurley | May 2026 |
| Zuckerberg’s comments | Mark Zuckerberg | Jul 2024 |
| Solaiman’s gradient paper | Irene Solaiman | Feb 2023 |
| Catalini’s economics paper | Christian Catalini | Aug 2026 |
The Regulatory Picture
Lawmakers have probed several companies over using Chinese models. DoorDash, Airbnb, and Anysphere/Cursor were questioned by CNBC, Bloomberg, and Semafor. Reuters reported on Apple.
Western companies have shifted from American closed labs to Chinese open models to save costs.
Key Dates
- List last updated: 13 Sep. 2026
- Gurley’s paper: “From Open Source Software to Open Source Strategy” (May 2026)
- Zuckerberg’s comments: “Open Source AI is the Path Forward” (Jul 2024)
- Solaiman’s paper: “The Gradient of Generative AI Release: Methods and Considerations” (Feb 2023)
- Catalini’s paper: “Some Simple Economics of Open versus Closed AI” (Aug 2026)
- ATOM Report: “ATOM Report” (Apr 2026)
What Readers Should Take Away
This is a remarkable resource. It is the work of a single person who has gathered a decade’s worth of research into one place, and made it freely available to anyone who wants to get up to speed.
The list rewards patience. Reading through it in order is a crash course in the economics, the safety debates, the geopolitical stakes, and the actual adoption numbers.
The collection is organized around the U.S. and China, with Lambert and Xu’s work on Chinese advantages forming a significant portion of the geopolitical section.
The papers are not dated. They range from 2023 to August 2026, and the list itself was updated as recently as September 13. The compiler has committed to updating it over time, which means the collection will continue to grow as new research appears.
For anyone who wants to understand the state of open models today, this is the place to start. The compilers have done the work of gathering the field’s best writing into one place, and they have asked for help adding to it.
Get the Notebook.
The day's best stories and every fresh verdict, in plain English, in your inbox by seven. One email a day, no more.