DeepSeek V3 0324

DeepSeek has released a new version of its model, DeepSeek V3 0324. This model is a refinement of the existing 671B-parameter Mixture-of-Experts (MoE) model, with 37B activated per token, and is available on the official DeepSeek platforms (website, app, mini-program) and Hugging Face analyticsvidhya.com. Early testers have reported significant improvements over the previous version, with one tester stating that it is now the best non-reasoning model, surpassing Sonnet 3.5 venturebeat.com. The technical report and weights for DeepSeek V3 0324 are accessible under the MIT license analyticsvidhya.com. DeepSeek, the company behind the model, was founded in 2023 huggingface.co.