DeepSeek Publishes V4.1 Flash Weights on Hugging Face With Integrated Vision Components
Sign in✓ Signed in
DeepSeek has published DeepSeek-V4.1-Flash on Hugging Face, making its latest experimental model available as MIT-licensed open weights. The repository includes a causal language model, a vision encoder and aligner, and reference inference code for the model’s multimodal and sparse-attention components.
Continue reading
The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in✓ Signed in — this article isn’t included in your current plan.
Unlocking the full article...