It's not clear which v4 flash they used. There are 3 different releases for that one.

Edit: I think that's the answer. This page lists the score as 53.3% which is close to the 54.4% reported here https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-0731 for the 0731 update. The original from April was called preview, doesn't list the score, but in the comparison with the update it scores 7.3%.

That makes sense. I wish they run it on some model released before deepswe release like gemini-3.5-flash.

DS flash also has V4.1 update and supposedly achieves 77.0 on deepswe v1.1

https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017