# Is DeepSeek V4 Pro Thinking SOTA?

**No.** #7 on LMArena Text (style-controlled, math); missing Humanity's Last Exam and FrontierMath

As of Oct 14, 2026, DeepSeek V4 Pro Thinking from DeepSeek. #7 on LMArena Text (style-controlled, math); missing Humanity's Last Exam and FrontierMath #7 on LMArena Text (style-controlled, math); missing Humanity's Last Exam and FrontierMath

As of Oct 14, 2026. Methodology v1.0. Computed automatically.

- Math: INSUFFICIENT_DATA — #7 on LMArena Text (style-controlled, math); missing Humanity's Last Exam and FrontierMath