Rayko

Midjourney’s aesthetic remains unmatched to this day.

DeepSeek 7 Quick Takeways: The recently viral fundraising remarks from DeepSeek founder Liang Wenfeng have now been removed across the Chinese internet. Before they disappeared, I read through them several times. Here are my main takeaways: My two strongest impressions were restraint and focus. MinMax also emphasizes AGI as its vision, but DeepSeek feels more credible to me because its actions seem more consistent with its words. The model is the core. Building lower-level products on top of higher-level technology is essentially a form of asymmetric advantage. This idea is both right and wrong. The capital market has recently cooled toward AI applications, partly because there still has not been a truly outstanding application that convincingly answers this question. The next generation of models may be able to improve through self-learning, and may even be capable of building better models themselves. DeepSeek’s current focus is on agents, especially coding agents. DeepSeek is genuinely cheap, but it claims this is not a price war. I find that argument slightly unconvincing. If DeepSeek’s coding capabilities were significantly ahead of everyone else, there is no way it would still be priced at the current level. The compute demand alone would be difficult to sustain. Open source is part of DeepSeek’s competitive strategy. Without open-sourcing its models, DeepSeek would not have achieved its current level of influence, nor would it have formed partnerships with major companies such as Tencent. To some extent, DeepSeek’s open-source strategy has genuinely accelerated the development of large language models. We can see DeepSeek’s restraint in many areas. It has not chased every trend by launching numerous apps and products, and it has not rushed into aggressive commercialization. I believe this is meant to preserve focus. Everything serves one ultimate goal: AGI. DeepSeek’s biggest challenge right now is team stability. Other major companies are willing to spend heavily to recruit its people. The situation appears to have improved somewhat following the latest fundraising round. DeepSeek comes across as humble and open. Its employees reportedly do not have strict KPIs, and overtime is not encouraged—perhaps that is also one reason its model updates have been relatively slow. It is difficult not to feel positive about a company with this kind of vision and posture. I only hope DeepSeek can continue to align its actions with its words. I am looking forward to its next-generation model—and hopefully, another true “DeepSeek moment.”

We just finished @hf0 Demo Day! We’re building Manyverse, turning AI video and world models into playable interactive worlds. Huge thanks to the HF0 team🙏, the founders I met, and everyone who stopped by our booth to try the demo and share feedback. I’m Ray, CEO & Co-founder of Manyverse. Excited to keep exploring the future of AI-native games. DMs open if you’d like to chat.

GTA 6 is almost here. https://t.co/cdlD35Nkev

Introducing Manyverse.@ManyVerse_ The platform that turns AI videos into playable game worlds. Watch the first trailer. The Manyverse is opening. https://t.co/lSqFFFiOJx