Shanghai AI Lab Shipped an Agent Model With No Announcement
Shanghai AI Lab dropped a 744B agentic model under MIT license with no blog, no paper, no PR. The strategy matters more than the model.
Tag archive
Shanghai AI Lab dropped a 744B agentic model under MIT license with no blog, no paper, no PR. The strategy matters more than the model.
US agencies accused six Chinese AI labs of systematic knowledge theft through industrial-scale distillation. The real story: frontier models have no border controls.
LG released K-EXAONE 2.0, a 750-billion-parameter open-source model, as South Korea's government-backed AI race enters its decisive phase.
Palo Alto Networks documented a Zhuhai-based actor weaponizing an open-source agent framework to automate reconnaissance and exploitation at scale.
DeepSeek releases a multimodal version of V4-Flash, pricing visual tokens cheap enough to make agent workflows viable.
OpenAI, Anthropic and Google share threat intelligence via the Frontier Model Forum to stop adversarial distillation. What it means for AI developers in 2026.
Moonshot's 2.8T model goes open-weight today, but infrastructure costs mean \"open\" is still a club. And there's a hallucination problem nobody mentioned.
Xiaomi deployed MiMo-V2-Pro to OpenRouter under a fake codename, let it top the charts for a week, then revealed who built it. The strategy worked perfectly.
Comparing Top Chinese AI Models: A Developer's Guide to DeepSeek, Kimi, and Other LLM...
Anthropic accused Alibaba-linked operators of running 28.8 million Claude interactions through 25,000 fake accounts to harvest model capabilities for Qwen.
GLM-5.1, MiniMax M2.7, Kimi K2.6, and DeepSeek V4 landed in 18 days, all frontier-competitive on coding benchmarks, all priced at a fraction of Claude Opus 4.7.
DeepSeek vs Western LLMs: A Developer's Guide to Chinese AI Models in 2024 In the rapidly...