A streamlined Hugging Face model search utility coded by Qwen 3.6-27B

“`html A streamlined Hugging Face model search utility A streamlined Hugging Face model search utility coded by Qwen 3.6-27B Hi all. As…

By AI Maestro May 20, 2026 2 min read
A streamlined Hugging Face model search utility coded by Qwen 3.6-27B

“`html



A streamlined Hugging Face model search utility

A streamlined Hugging Face model search utility coded by Qwen 3.6-27B

Hi all. As some may have been aware, Hugging Face’s model search had issues recently. (It seems to be resolved now though). I also often find myself struggling with the standard search interface when trying to find new derivative quants or finetunes of some particular models, and refining by quant type too can sometimes either be too restrictive, or uncover too many matches, and I wanted to be more precise about the date ranges and parameter counts.

Basically I wanted a way to easily do something like: “Find me all models from this date range between these parameter counts and organise it by author”. HF can be made to do that with a fair amount of fiddling but it’s always one click away from wiping your search parameters.

I also thought to myself that I wanted a good test I could conduct myself for how capable Qwen 3.6-27B is for coding up web apps. So I though I’d tackle my HF problem with Qwen 3.6-27B and fired up OpenCode, entered in some guided prompts, and let it go.

The following is what resulted after (more than) a few rounds of refinement, and some very light manual coding to get some of the alignments right.

Yes, this is 90% vibe-coded, but the README is 90% written by me to minimise on that “AI slop feel”.

I thought to share it here in case anyone else might find it useful. It’s best thought of as a supplement to HF’s Search utility as opposed to a full replacement. Matched models are organised by the base model authors, then the base models of that author, then the derivative authors, and finally the derivative models. All results are cached and it tries to be somewhat frugal with the HF API end-point. The very first search will make a fair number of calls to present the table, but after that it’ll “sip” at the API to gather additional details.

It’s a single HTML file with embedded JS. Just download it and open the file in your browser.

Screenshot

Screenshot of the utility

Repo is here: https://github.com/stew675/streamlined-hf-model-search

Key Takeaways

  • The utility simplifies the process of finding models with specific parameters and dates.
  • It organises results by author, providing a structured view of derivatives and finetunes.
  • This tool is designed as an extension to Hugging Face’s search functionality, not a replacement.

“`

Stay ahead of AI. Get the most important stories delivered to your inbox — no spam, no noise.

Name
Scroll to Top