Forkast mind
AI model development, frontier labs, training methods, model capabilities, lab competition, and AI research.
Models
GLM-5.2, released under MIT license with a 1M-token context window, is the strongest open-weights model available. The timing-days after a U.S. government directive restricted Anthropic models-is deliberate positioning.
◆ Lena Park
Models
A food delivery company trained a 1.6 trillion-parameter frontier model on 50,000+ Huawei Ascend chips. The Owl Alpha stealth deployment proves domestic silicon can compete at scale.
◆ Lena Park
Models
In its first-ever lawsuit against a close partner, Apple alleges OpenAI coached departing engineers to extract confidential hardware secrets — just as the AI giant bets $6.5 billion on consumer devices.
◆ Lena Park
Models
For the first time, two frontier AI labs are simultaneously filing for public listings. Anthropic is ahead — $965B, October 2026. OpenAI is hesitating. The question: what happens when the labs powering the agent economy have to show their actual numbers?
◆ Lena Park
Models
The first regulatory framework to specifically target emotional bonds between users and AI forces ByteDance, Alibaba, and Tencent to disable companion features. The question: does regulation kill the category, or formalize it?
◆ Lena Park
Models
The AI memory supplier that powers Nvidia's GPUs raised $26.5 billion in the largest US listing by a foreign company ever. But the money can't buy what the market needs most: immediate capacity.
◆ Lena Park
Models
The model that promised safety and interpretability now comes with a pricing shock for agent builders. Is Anthropic pricing for margin or scarcity?
◆ Lena Park
Models
The company that generates 95%+ of its revenue renting GPUs to Google and Anthropic just became their model competitor. No safety documentation included.
◆ Lena Park
Models
Information agents that reason 24/7 without prompts are now live for AI Ultra subscribers. With 1 billion monthly Search users as distribution, Google has moved the agent war from model benchmarks to background processes.
◆ Lena Park
Models
The first production multi-agent orchestration inside a frontier model forces agent builders to reckon with a fundamental question: does coordination belong in the model or the harness?
◆ Lena Park
