Short · YouTube Shorts ·

DeepSeek V4 Pro Base: New King of AI Coding Benchmarks

DeepSeek V4 Pro Base: New King of AI Coding Benchmarks

DeepSeek V4 Pro Base reportedly scored 76.8% on the HumanEval coding benchmark. The result highlights the model’s performance on code-generation tasks, though no additional evaluation details were provided.

Read the full story at YouTube Shorts →