We have taken another step towards the future. A few seconds of AI video has evolved into a real-time rendered world. The first real-time world model PixVerse R1 has been released. Although it's just a very, very early version, it is definitely worth experiencing.

Waking up early brought good news. We made it to the Final 10 of the NVIDIA 2025 Startup Accelerator Program (ranked first, although the ranking is not in order). In 2026, we will go to the U.S. to attend NVIDIA GTC and meet Jensen Huang. The accelerator has been very enthusiastic and helpful to us; I never expected to make it to the Top 10 when we first joined, it feels like a dream 😂.

Today I suddenly realized that Google can create good large models, but Meta and Apple cannot. The technology for large models has actually become a strong technical barrier. It's just that Google is particularly strong and has broken through the barrier.