Submit Your Setup
Add your local LLM to the community matrix β no account, no server.
The 30-second version: The matrix runs on community data. If you run a local LLM setup, share it β no account, no server, no guesswork. Fill in the form and we'll assemble a ready-to-post [COMMUNITY-SUBMISSION] block. You can copy it for the forum thread or open a pre-filled GitHub issue, whichever you prefer. Your name is always credited.
Why submit?
Every row in the Matrix is a real setup someone is actually running. Your data helps answer the site's core question β what should I run, on what I have, for what I need? β for the next person with your hardware. A 30-second submission is what turns a one-off report into a verified, citable data point.
How it works
- Fill in the form. Only three fields are required β your VRAM tier, hardware, and model. That's the 30 seconds. Everything else (backend, quant, speed, use case, notes) is optional β but the more you share, the more useful your entry is.
- Pick your path. Copy the block straight to the forum thread (the original home of the data), or open a pre-filled GitHub issue.
- We integrate it. Submissions are reviewed weekly and parsed into the dataset β your entry appears in the Matrix and the forum index, credited to you.
Not sure about a field? Leave the optional ones blank. A quick report with just your GPU, model, and "it works" is genuinely valuable β it corroborates other submissions and helps the matrix converge on the truth.
Build your submission
Preview β copy this block
What happens next
Reviewed weekly. Submissions are read, questions are asked in-thread when a number looks off, and valid entries are parsed into the dataset.
It appears in the index. Your entry shows up in the Matrix and the forum index, credited to you.
Corroboration promotes it. When an entry is confirmed by 2+ independent sources β or tested by the maintainer β it's promoted to the master matrix. No submission is too small to matter.