Qwen4-Max leaks at 2.4T MoE with 1M context; Tencent launches $0.07 Hy4 Flash
testingcatalog · x · 2026-11-16
TestingCatalog's Nov 16 AI brief (self-described as Grok-assisted, unconfirmed leaks):
- Alibaba: Qwen4-Max out — a 2.4T MoE model with native vision and 1M context at $1.50/$4.50 per 1M tokens.
- Tencent: Hy4 Flash launched, a 1M-context multimodal model at $0.07/$0.281 per 1M tokens, 50% off through Nov 30.
- OpenAI: testing a new desktop ChatGPT web UI with cards for content creation and activities.
- Google: Antigravity doubling context windows for Gemini 4 Argon/Carbon/Hydrogen; a Comic artifact with style examples is coming to Gemini Notebook.
- Meta: Muse Hatch waitlist opened in the US.
- Apple: per Bloomberg, Siri AI app integrations will headline next year's smart glasses.
- NVIDIA: per FT, Jensen Huang reportedly considering returning as CEO.
The quoted Nov 14 brief also covered ChatGPT Wallet rollout for US Pro users, Antigravity's Dynamic Workflows (/run-workflow scripts orchestrating subagents), and an "Omni LoRA tuning" spec for Gemini Omni video fine-tuning in Vertex AI.
Related event: Alibaba Releases Qwen4-Max, Top Chinese Model Across Arenas(47 posts)→
More from Companies & People
- Study of 95 startups: 56 in 100 news mentions are just passing references — YvesMulkers · 2026-11-16
- Ex-Anthropic safety researcher Jacob Coxon unsure quitting was right, may return — SandraWachter5 · 2026-11-16
- "You pay for it and you're still the product": the case for self-hosted models — bendee983 · 2026-11-16
- Unverified Huang ouster conspiracy theory splits OpenAI and Anthropic employees' reactions — bgmshana · 2026-11-16
- Ryanair cuts disruption video production time up to 90% with Synthesia AI video — synthesiaIO · 2026-11-16
- Commentator argues Nvidia's Jensen Huang is the biggest obstacle to Anthropic's regulatory ambitions — cpheinrich · 2026-11-16