Alibaba's Qwen 4 already training as company targets 5-10 trillion parameter models
emmanuelvivier · x · 2026-09-23
- Alibaba is targeting Qwen models at 5 to 10 trillion parameters, with Qwen 4 already in training—a scale ambition well beyond current frontier models.
- Context from the thread: Alibaba also unveiled the Zhenwu V900 chip, billed as China's most powerful AI chip, designed for 500,000-card clusters.
More from Models
- Same prompt, two new models: Claude Opus 5.5 vs GPT-6 Sol launch video comparison — CurieuxExplorer · 2026-09-23
- Alibaba Launches Qwen Intelligence With 3 SOTA Mobile Agents, 90% End-to-End Success Rate — Alibaba_Qwen · 2026-09-23
- Google slammed for model drought: no flagship since Gemini 3.1 Pro in February — CtrlAltDwayne · 2026-09-23
- Guardrails that block all PoC generation flood vendors with hallucinated bug reports, says researcher — dyn___ · 2026-09-23
- DrivingBench lets ChatGPT, Claude, and Grok drive a real Toyota Corolla; only one finishes the course — pwlot · 2026-09-23
- Users grumble as ChatGPT's 'Latest' model selector hides which model you're actually using — koltregaskes · 2026-09-23