DeepSeek Releases Faster V4.1 Model with Image Capabilities

Chinese AI lab DeepSeek has introduced a new open-source model named V4.1-Flash while retiring its current top tier system, V4-Pro. As reported by TNW, starting September 14, all user traffic directed to the older model will automatically route to V4.1-Flash at lower rates. The new system handles both text and visual inputs, utilizing a selective processing design that keeps memory requirements low.

For everyday consumers and software developers, this update makes running automated digital tools significantly cheaper. According to Bloomberg Intelligence, the shift cuts operating costs by up to 32 percent, applying fresh financial pressure on competing tech firms. Because DeepSeek released the system under a permissive license, developers worldwide can freely download and run the software on their own infrastructure.

However, several uncertainties remain regarding its safety and reliability. DeepSeek's self-reported benchmarks show strong performance in coding tasks, but TNW noted that the model trails top American systems in advanced academic reasoning and complex image analysis. As foreign open-source models become increasingly capable and affordable, organizations will need to carefully balance these cost benefits against ongoing data privacy and compliance considerations.