Everyone’s watching the race to the best AI model.
Model performance is becoming a commodity. What surrounds it is what will matter most.
This is the shift OpenAI and Anthropic have already started to make and it affects every builder using AI.
As open models get faster,
Everyone’s watching the race to the best AI model. Model performance is becoming a commodity. What surrounds it is what will matter most. This is the shift OpenAI and Anthropic have already started to make and it affects every builder using AI. As open models get faster,
Down the line, the best founders will likely not be comparing foundation model capabilities, but rather the “build versus buy” decision to create scaffolding around large models (open or closed) for specific needs. The ability to point a model at a specific use case will become the most important problem in the value chain of AI, so foundation mode... See more
The Real Value of AI Isn’t General Intelligence
As the LLM market structure stabilizes, the next frontier is now emerging. The focus is shifting to the development and scaling of the reasoning layer, where “System 2” thinking takes precedence. Inspired by models like AlphaGo, this layer aims to endow AI systems with deliberate reasoning, problem-solving and cognitive operations at inference time... See more
Sonya Huang • Generative AI’s Act O1
过去一年，我一直跟大家讲，deepseek、moonshot、SRAM推理三家，都是可能挑战OpenAI的存在。
moonshot就不说了，他们解决的核心问题是context window，这是用户解决复杂问题的核心需求，moonshot最开始给出了非常惊艳的答案——可惜后来别人也追上来了；
deepseek v2走了mixture of experts的路， 这条路Mixtral 8x7B已经帮他们探明了，就是用一个零头的training cost去完成training，就是省钱；另一个是multi head latent attention，能让deepseek早期用一堆烂显卡以极低的成本提供inference服务（0.28刀提供1M参数）；
SRAM推理三家（Groq、Samba... See more
