A global sell-off in AI and chip stocks stabilized after midterm results and data-center fears rattled markets, even as labs kept shipping at a relentless pace. OpenAI, Google, Meta, Tencent, and Thinking Machines all rolled out major releases this week, while Anthropic's IPO timing came into question. The tension between breakneck model progress and investor caution is now the industry's defining storyline.

OpenAI launches GPT-6 Cyber after it found 81 Firefox flaws and 948 Linux exploits OpenAI unveiled GPT-6 Cyber, a cybersecurity-focused AI model that discovered 81 previously unknown high-severity Firefox vulnerabilities and developed working exploits for 948 distinct Linux flaws. Why it matters: An AI model autonomously finding and exploiting vulnerabilities at this scale reshapes both defensive security and the threat landscape.

AI stocks stabilize after global sell-off tied to midterms and data-center fears AI and chip stocks tumbled worldwide after Democratic midterm victories were read as a rebuke of US data-center expansion, with TSMC down 9.6%. Markets later halted the sell-off as Wall Street trimmed weekly losses and a US jobs report beat expectations. Why it matters: The sell-off shows political and infrastructure concerns can now move AI markets as much as product news.

Anthropic IPO timing in doubt as AI backlash spooks investors Potential investors are questioning whether Anthropic can proceed with its planned IPO roadshow next week, with doubts raised about timing amid the AI backlash and market sell-off. Prediction-market odds for the listing have fallen. Why it matters: A delayed or derailed Anthropic IPO would signal how quickly sentiment can turn on even the strongest AI companies.

Google makes Gemini Omni 1.2 Flash generally available with 15-second clips Google released Gemini Omni 1.2 Flash on November 5, making its video generation and editing model generally available via the Gemini API and Gemini for business, with clips up to 15 seconds, native audio, and roughly 30% lower cost. Why it matters: Cheaper, generally available video generation pushes AI video from demo to production tool for developers and businesses.

Thinking Machines ships Inkling 2 Small amid reported $5B raise at $40B valuation Mira Murati's Thinking Machines Lab released Inkling 2 Small, a 594B-parameter MoE with 17B active parameters, claiming the top spot among western open-weight models. The startup is reportedly raising $5B led by Accel at a $40B pre-money valuation. Why it matters: A top-ranked western open-weight model plus a massive raise would make Thinking Machines a serious counterweight to the closed-model giants.

Tencent launches Hy4 multimodal model with vision and computer use Tencent released Hy4 on November 4, a 770-billion-parameter mixture-of-experts model with about 49 billion parameters activated per query, adding visual understanding and computer use. API promo pricing is $0.417/$1.251 per million tokens. Why it matters: Aggressive pricing on a frontier-class multimodal model intensifies Chinese labs' push into Western developer markets.

Meta adds real-time voice and expressive avatars to Muse app Meta announced on November 5 that real-time voice conversations and expressive avatars are rolling out in the Muse app in the US across mobile, web, and Mac, letting users talk to their AI agent instead of typing. Why it matters: Voice-first AI interaction at consumer scale is Meta's bid to make agents a daily habit rather than a chat box.

OpenAI launches Images 2.6 with animated GIF creation and faster API models OpenAI rolled out Images 2.6 across ChatGPT, Codex, and the API, adding animated GIF creation from prompts or photos, faster generation, and two new API models named Flare and Sunburst. Why it matters: Native animation and faster API models extend OpenAI's image lead into motion content for everyday users and developers.

METR finds Anthropic's AI improving twice as fast despite CEO's slowdown call Outside research group METR found Anthropic's models are improving capabilities roughly twice as fast as a year earlier, even as CEO Dario Amodei urged the industry to slow down so safety measures can catch up. Why it matters: The gap between safety rhetoric and measured capability growth is widening, raising hard questions about voluntary slowdowns.

DeepSeek-V4.1-Flash gains traction as local inference tooling matures Community reports show DeepSeek-V4.1-Flash running on consumer and older datacenter GPUs, with the Delta Inference Engine v0.93 improving quality and memory use. Software tuning has boosted throughput up to 60% on unchanged NVIDIA hardware. Why it matters: Frontier-adjacent models running on consumer hardware accelerate the shift toward local, low-cost AI inference.

Read online · Live feed · You get this because you subscribed at ainewsnow.io — unsubscribe anytime with the link below.