MarkTechPost·14d agoCognition Releases SWE-2: A Kimi K3 Post-Trained Coding Model That Matches Fable 5.1 on FrontierCode at 64% Lower Cost#benchmark#coding-agent#cognition 4 min1
TechCrunch · AI·15d agoKimi-maker Moonshot AI targets $2 billion in annual revenue#anthropic#china#claude-opus
Hacker News · AI·15d agoMoonshot serves Claude instead of Kimi and collects exchanges for model training#ai-industry#anthropic#claudeAI industry2
TechCrunch · AI·16d agoAnthropic details distillation campaigns from Alibaba, Moonshot AI, and DeepSeek#alibaba#anthropic#chain-of-thought 3 min3
BleepingComputer·17d ago highUS says Chinese firms extracted billions of tokens from frontier AI models#alibaba#cisa#deepseek in the wild 3 min
SecurityWeek·17d agoUS Agencies Warn China Is Systematically Extracting Frontier AI Capabilities#china#cisa#deepseek in the wild 3 min2
Help Net Security·17d ago highChinese AI firms are siphoning capabilities from American models, CISA warns#china#cisa#deepseek in the wild 4 min2
The Hacker News·17d agoU.S. Agencies Accuse China AI Firms of Distilling Claude, GPT, Gemini, and Grok#china#cisa#deepseek 5 min1
CyberScoop·18d agoFeds accuse China of ‘systematic’ distillation of U.S. AI models#ai-policy#china#cisa 4 min1
Hacker News · AI·18d agoKimi K3 (2.8T) at 1 token/s on a MacBook Pro, streamed from four SSDs#inference#kimi-k3#local-inferenceAI tools & infra