# Model selection

**URL:** <https://discuss.huggingface.co/t/model-selection/168440>\
**Category:** Models\
**Created:** [September 15, 2025, 11:48am UTC](https://discuss.huggingface.co/t/model-selection/168440 "2025-09-15T11:48:52Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![olucia](https://avatars.discourse-cdn.com/v4/letter/o/839c29/32.png) [@olucia](https://discuss.huggingface.co/u/olucia)\
**Post date:** [September 15, 2025, 11:48am UTC](https://discuss.huggingface.co/t/model-selection/168440/1 "2025-09-15T11:48:52Z")

</div>

Hi all,

I am a Product Manager looking at a model selection for a conversational UX. Reaching out for input from technical folks and anyone knowledgeable as to how to go about a model exploration and selection in general.

What should I consider? What is the best place to search for a model? And any further considerations?

Hugging Face hosts more than 2 million models to choose from, good documentation

I’ve been using AWS and GCP for labs and experiment apps too

GCP Model Garden

AWS Bedrock

Thank you for any suggestions you might be willing to share

---

<div class="post-metadata">

**Author:** ![John6666](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/john6666/32/27664_2.png) [@John6666](https://discuss.huggingface.co/u/John6666)\
**Post date:** [September 15, 2025, 12:23pm UTC](https://discuss.huggingface.co/t/model-selection/168440/2 "2025-09-15T12:23:09Z")

</div>

Ultimately, I think it’s best to actually run an LLM locally or in the cloud to get a feel for it before choosing one.

But anyway, there are a lot of models. Many people use leaderboards to narrow down their options. These are rankings sorted by benchmark scores. There are leaderboards for various themes based on benchmarks for different tasks.

Also, if you have a specific task in mind, it’s a good idea to look into well-regarded models and their derivative models on forums related to that task.

### Leaderboards

- [Leaderboards on Hugging Face Spaces](https://huggingface.co/spaces?sort=trending&search=leader)
- [The Big Benchmarks Collection](https://huggingface.co/collections/open-llm-leaderboard/the-big-benchmarks-collection-64faca6335a7fc7d4ffe974a)
- [SWE-bench Bash Only](https://www.swebench.com/bash-only.html)
- [Aider LLM Leaderboards](https://aider.chat/docs/leaderboards/)

---

<div class="post-metadata">

**Author:** ![aiflux](https://avatars.discourse-cdn.com/v4/letter/a/9de053/32.png) [@aiflux](https://discuss.huggingface.co/u/aiflux)\
**Post date:** [September 15, 2025, 1:40pm UTC](https://discuss.huggingface.co/t/model-selection/168440/3 "2025-09-15T13:40:50Z")

</div>

I generally recommend trying models on rented GPU’s first.

Leaderboards are great - but communities themselves always seem to have “favorite” models as well.

for instance:

- image gen models

- creative writing models

- coding models