DeepSeek Open-Sources DeepInfer Framework to Cut AI Deployment Costs

DeepSeek open-sources new tool to slash AI costs in ‘fight’ against closed-source rivals

DeepSeek has fully open-sourced an internally developed framework aimed at making artificial intelligence models cheaper and more efficient to deploy. The Chinese artificial intelligence start-up’s move comes as it describes the divide between open-source and closed-source AI as a battle between “light and darkness”, according to a senior researcher at the company.

Framework supports two chip platforms

The company released DeepInfer on Tuesday as an inference framework tailored to support its DeepSeek-V4.1-Flash model. The framework is designed to run the model on Nvidia’s SM100 Blackwell graphics processing units (GPUs) as well as Huawei Technologies’ Ascend 950 chips.

Inference refers to the serving and running of AI models. Frameworks such as DeepInfer help optimise that process, making the software layer used to operate models more efficient.

By open-sourcing DeepInfer, DeepSeek is making its internally developed framework available as part of its broader open-source approach. The company’s stated focus is reducing the cost of deploying AI models while improving efficiency, placing the release within its wider contest with closed-source rivals.

Source: South China Morning Post