# New models

**URL:** <https://community.sambanova.ai/t/new-models/567>\
**Category:** Discussion\
**Tags:** dev-interest, dev\
**Created:** [November 15, 2024, 10:07pm UTC](https://community.sambanova.ai/t/new-models/567 "2024-11-15T22:07:33Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![hello1](https://sea1.discourse-cdn.com/sambanova/user_avatar/community.sambanova.ai/hello1/32/2812_2.png) [@hello1](https://community.sambanova.ai/u/hello1)\
**Post date:** [November 15, 2024, 10:07pm UTC](https://community.sambanova.ai/t/new-models/567/1 "2024-11-15T22:07:33Z")

</div>

I want to ask if there are any plans for additional models.

I have a list of models I use now, and I’m confident using multiple models gives better results on the executions, especially for specialized agents.

Like Qwen 2.5 or Yi 1.5 or Hunyuan-Large

I have one additional question about this topic. Is there a chance of this happening in the commercial API, even if it’s not on the developer cloud?

---

<div class="post-metadata">

**Author:** ![hello1](https://sea1.discourse-cdn.com/sambanova/user_avatar/community.sambanova.ai/hello1/32/2812_2.png) [@hello1](https://community.sambanova.ai/u/hello1)\
**Post date:** [November 15, 2024, 10:19pm UTC](https://community.sambanova.ai/t/new-models/567/2 "2024-11-15T22:19:22Z")

</div>

One interesting fact that may justify why it is interesting:

**Why it matters:** Hunyuan-Large generally outperforms Llama 405B, achieving the performance of a 405 billion parameter model while computing only 52 billion parameters. That’s a significantly lower processing requirement, and the model is free for many purposes.

The other fact is that using radically differently trained models in a react model creates better results based on my experience.