AI Models

🎧 Prefer to listen? Your browser does not support the audio element. ByteDance is training an AI model with as many as 10 trillion parameters, and if you ship anything on top of AI APIs, this changes your economics even if you never send it a single request. For scale, that’s roughly three times bigger than Kimi K3, the largest Chinese model released so far, and it edges past most estimates of what Anthropic has in training. The Financial Times broke the story in August 2025, and the part I keep coming back to isn’t the parameter count. It’s what a third serious player does to the price of the tokens I buy every month. ...

🎧 Prefer to listen? Your browser does not support the audio element. Dario Amodei, CEO of Anthropic, testified to Congress in May 2025 that Chinese labs are running “industrial-scale distillation” on American AI models right now. He claims this process—using a top-tier model’s outputs to train a smaller, cheaper one—is being used by military-linked researchers to build defense systems. His push for controls on computing power, not just model usage, marks a new kind of choke point for the AI industry. ...

🎧 Prefer to listen? Your browser does not support the audio element. Six months ago, picking an AI tool felt like choosing sides in a rap beef. Anthropic or OpenAI? Claude or ChatGPT? Every thread on X was people arguing about which one was “better” like there was a single answer. I spent way too much time reading those threads instead of actually building things. ...

🎧 Prefer to listen? Your browser does not support the audio element. Something shifted in AI last week, and most people missed it. On June 17, Z.ai, a Beijing company spun out of Tsinghua University, released an open-source model called GLM-5.2. It beat GPT-5.5 on multiple coding benchmarks at roughly one-sixth the cost: 62.1 vs 58.6 on SWE-bench Pro, and 74.4% vs 72.6% on FrontierSWE, while shipping under the MIT license with a 1 million token context window. ...