DeepSeek updates V4.1 Flash, promising faster tasks at the same price

DeepSeek has updated V4.1 Flash, the AI model it introduced last month, with a release called V4.1-Flash-1017. The Chinese company says it performs better on tests for coding, reasoning, and software-agent work—tools that take several steps to complete a task—while finishing jobs faster and generating less text. That matters because every extra step and generated word adds computing time and API expense; DeepSeek says existing users receive the update through the same website, app, and developer model name, with no price change. DeepSeek's performance and cost claims still need independent testing.