Chinese startup DeepSeek has released an updated version of its R1 reasoning AI model on the developer platform Hugging Face after announcing it in a WeChat message Wednesday morning.
The updated R1, which is under a permissive MIT license, meaning it can be used commercially, is a “minor” upgrade, according to DeepSeek’s WeChat announcement. The Hugging Face repository doesn’t contain a description of the model — only configuration files and weights, the internal components of a model that guide its behavior.
Weighing in at 685 billion parameters in size, the updated R1 is quite hefty. (“Parameters” is synonymous with “weights.”) Without modification, the model likely can’t run on consumer-grade hardware.
DeepSeek rose to prominence earlier this year following the release of R1, which gave models from OpenAI a run for their money. The startup has raised the ire of some regulators stateside, who argue that DeepSeek’s technology poses a national security risk.