AI | DeepSeek Web Version Upgrade: Expert Mode Now Available for Handling Complex Issues

robot
Abstract generation in progress

China-made large model DeepSeek’s web version went live with two interactive modes—“Quick Mode” and “Expert Mode”—last night, supporting concise search and long-range, complex problem handling.

According to tests by Sina Technology, the “Quick Mode” introduced in this upgrade is closer to the earlier version of DeepSeek with direct naming changes. By contrast, “Expert Mode” adds professional knowledge consultation capabilities in fields such as programming, law, and medicine.

In handling mathematical problems, Expert Mode performs better than Quick Mode. For example, when asked whether a 7-meter-long sugarcane can pass through a door that is 2 meters high and 1 meter wide, Quick Mode responds that it cannot pass, but Expert Mode gives a deterministic answer that it can pass, along with a vivid analogy that the sugarcane’s cross-section is small enough to be pierced through like a spear.

In the view of industry insiders, the newly launched Expert Mode likely uses a larger, stronger model, possibly a configuration of the official DeepSeek V4 release. However, since it currently does not support document uploads and lacks multimodal capabilities, beyond improvements to long-range and complex question-answering, more capabilities still await further release from DeepSeek.

View Original
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
  • Reward
  • Comment
  • Repost
  • Share
Comment
Add a comment
Add a comment
No comments
  • Pin