Open-source AI by the numbers
Figures from the Forkopolis catalog, recalculated every week from public data. Last updated November 11, 2026.
- Forkopolis tracks 57 free, open-source AI tools and 35 paid AI services.
- 70% of the open-source tools (40 of 57) run entirely on your own computer.
- 68% (39 of 57) work without a dedicated graphics card.
- 63% (36 of 57) can be used on a computer with 8 GB of memory and no graphics card.
- 23% (13 of 57) have an easy installer.
- The most common license is MIT, used by 24 tools.
Tools by category
Most common licenses
| License | Tools |
|---|
| MIT | 24 |
|---|
| Apache-2.0 | 16 |
|---|
| AGPL-3.0 | 6 |
|---|
| GPL-3.0 | 3 |
|---|
| MPL-2.0 | 1 |
|---|
Most starred projects
| Project | GitHub stars | Score |
|---|
| OpenCode | 226,151 | 4.9 |
|---|
| MarkItDown | 194,948 | 4.9 |
|---|
| Ollama | 186,679 | 4.8 |
|---|
| Stable Diffusion web UI | 165,287 | 4.6 |
|---|
| Dify | 160,992 | 4.5 |
|---|
| Open WebUI | 158,409 | 4.5 |
|---|
| Langflow | 156,547 | 4.8 |
|---|
| ComfyUI | 146,362 | 4.8 |
|---|
| llama.cpp | 134,152 | 4.8 |
|---|
| Browser Use | 120,733 | 4.7 |
|---|
Where these figures come from
Stars, licenses, and update dates come from each project's public GitHub repository. Hardware needs and setup effort are Forkopolis estimates. You are welcome to quote these figures with a link to this page. The full catalog is available as a data file.