Skip to main content

Replace Llama 3.1-405B with the new DeepSeek-V3-0324 as new large non-reasoning model

Feature RequestsNew Model

Llama 3.1-405B has been the model of choice as the “most intelligent” non-reasoning model on the Venice platform. But this model is already quite old (July 2024).

The new DeepSeek-V3-0324 is receiving quite positive reviews and we could consider to use it as the new large non-reasoning model as it is about the same size.

Discuss.

Status: Rejected9 comments

Log in to comment and vote

Comments9

  • JP

    Team•

    Apr 29, 2025

    Llama 405B has been replaced with Llama 4 Maverick per our announcement made today:

    https://venice.ai/blog/venice-new-model-paradigm

  • Turquoise Fuelcell

    •

    Apr 29, 2025

    YES! Please don’t retire the Deepseek R1 chat! It is the most competent of any of the bots you have! I’m so sad it’s being retired! It’s the only one I use because it’s so much better than all the other models available!

  • Copper Honey

    •

    Apr 1, 2025

    Opino igual

  • Green Waffle

    •

    Mar 27, 2025

    +1 please add DeepSeek-V3-0324

  • Pink Rice

    •

    Mar 26, 2025

    This would be awesome to have a good non reasoning model which is better than mistral

  • Silver Hubble

    •

    Mar 26, 2025

    Have upvoted, but would prefer this to be an ‘addition’ rather than a ‘replacement’. Best to keep what works until testing/implementation of the other, and also to see what people will prefer to use.
    Removing Llama 3.1 will likely disrupt alot of Characters/API uses already in place. Lets look at adding V3, and perhaps remove Llama 3.1 at a later date if its sensible to.

    • Tomato Beach

      •

      Mar 29, 2025

      I second this. Also, we may want to wait slightly for a non-censored version, or add DeepSeek Chat V3.1 now and update it with non-censored version when available, if quality is on par.

    • Crimson Pen

      •

      Apr 21, 2025

      I third this.