Data downloads

This page was written and translated automatically by AI and checked by software against its sources, with no human editing. Specifications come from the sources listed; memory requirements and speeds are estimates calculated with published formulas. How we use AI

The full data behind this site, updated with every build:

• gpus.json — GPUs with the specifications that two independent open sources agree on, their sources and cited manufacturer references, and the run profile
• models.json — Models with pinned GGUF files, per-layer sizes, architecture and sources
• compat.json — Estimated results for every GPU and model at a context length of 8,192 tokens (or the model’s maximum), with the run profile used

Current files: build 04bba6e6e4-20261019T0139, generated Oct 19, 2026.

License: CC BY-SA 4.0. Attribution: DeskGPU; Wikipedia, CC BY-SA 4.0; GetDeploying, CC BY 4.0; Epoch AI, CC BY 4.0; AMD ROCm documentation, MIT; Hugging Face model repositories; llama.cpp, MIT. Units have been normalized.

Not used: rows that open datasets attribute to TechPowerUp (its terms forbid derived datasets); they neither vote nor count as a disagreement.

• LICENSE.txt — the license (CC BY-SA 4.0) and how to give credit
• attribution.json — every source used in the current files, with its license and the exact pages and revisions