DeepSeek’s latest experimental AI model, called DeepSeek V3.2 Exp, is here. It is built on the company’s previous V3.1-Terminus capabilities. The new model introduces DeepSeek Sparse Attention (DSA), which is designed to speed up training and inference, particularly for long context tasks without compromising on output quality.
DSA offers fine-grained sparse attention, which reduces compute costs and enhances performance when handling large amounts of data. Early benchmarks indicate that V3.2-Exp performs on par with V3.1-Terminus, making it a promising upgrade for developers and businesses relying on long-context AI tasks.
The update is now available across App, Web, and API platforms. Alongside the model launch, DeepSeek has cut API prices by over 50%, making it more accessible to developers and teams. For those who want to test and compare results, V3.1-Terminus will remain accessible via a temporary API until October 15, 2025, 15:59 UTC.
With this launch, DeepSeek is focusing on efficiency and affordability, giving users faster performance without sacrificing accuracy. However, if you have multiple AI API keys and looking for a way to streamline your AI workflow without compromising on efficiency and affordability then tools like Geekflare Connect can be useful. It allows you to link APIs from DeepSeek, OpenAI, Google, Anthropic, and more in a single workspace, compare model outputs side-by-side, and track usage.
