"AI Disruption" Publication 5000 Subscriptions 20% Discount Offer Link.
Last night, DeepSeek V3 received an unexpected update, upgrading to the "DeepSeek-V3-0324" version.
Currently, the new version is available for download and deployment on Hugging Face.
DeepSeek-V3-0324 now supports the more permissive MIT open-source license.
How It Works
This 641GB model made its debut on Hugging Face with an empty README file and published weights, once again with almost no promotion.
DeepSeek-V3-0324 operates in a 4-bit mode, achieving speeds exceeding 20 tokens per second on a 512GB M3 Ultra equipped with mlx-lm.
