OpenAI preps new Reasoning Model for next week

PLUS: Runway Gen-4.5 tops text-to-video benchmark & DeepSeek V3.2 rivals GPT-5 at 30x lower cost. Kuaishou unveils Kling O1 video platform, Apple AI chief steps down.

In today’s agenda:

1️⃣ OpenAI enters “Code Red” mode preparing new reasoning model for next week's launch to counter Gemini 3's market share gains

2️⃣ Runway releases Gen-4.5, topping the Artificial Analysis Text-to-Video leaderboard and surpassing Veo 3 and Sora 2 Pro with realistic physics and fluid motion

3️⃣ DeepSeek launches V3.2 and V3.2-Speciale models matching GPT-5 performance at $0.28 per million tokens, nearly 30 times cheaper than competitors

  • Kuaishou launches Kling O1, unified AI video system combining generation and editing capabilities in single integrated workflow

  • Apple's AI chief John Giannandrea stepped down following repeated Siri delays

H-FARM EVENTS ON AI

AI UNPLUGGED²

After last year’s great success, AI UNPLUGGED is back with a new edition, created in collaboration between H-FARM AI and H-FARM Business School, and supported by CISCO. This time, we invite you to get hands-on and unlock real-world applications and potential of AI for Business.

📅 12.12.2025 – 6:00 PM
📍 H-FARM Campus

Special guest: Pier Luigi Pisa, journalist at la Repubblica and Italian Tech, one of the most active Italian voices in the debate on the ethics and implications of artificial intelligence, and author of the book “Valeva la pena tentare”.

With contributions from:
Alessandro Petrillo – CEO, H-FARM Business School
Diego Pizzocaro – CEO, H-FARM AI
Enrico Miolo – CISCO
Michele Dalmazzoni – CISCO

AI is not the future: it’s the present.

🔤 Please note: the event will be held in Italian - Limited seats available

MAIN AI UPDATES / 2nd December 2025

🚨 OpenAI preps new reasoning model for next week 🚨
Company enters "Code Red" mode as Gemini gains market share.

OpenAI is mobilizing for the imminent launch of a new reasoning model that CEO Sam Altman claims outperforms Google's upcoming Gemini 3 in internal benchmarks, with release targeted for as early as next week. The accelerated push comes amid internal "Code Red" concerns as Google's Gemini gains market share and reclaims benchmark leadership positions. To concentrate resources on this critical launch, OpenAI is temporarily pausing multiple initiatives including advertising campaigns, new autonomous agent development, and its "Pulse" news briefing service. The company is redirecting engineering efforts toward core ChatGPT improvements and urgent enhancements to its image generation capabilities, signaling the strategic importance of maintaining competitive positioning against Google's advancing AI systems.

🎬 Runway Gen-4.5 tops text-to-video benchmark 🎬
New model surpasses Veo 3 and Sora 2 Pro with cinematic physics.

Runway unveiled Gen-4.5, a new AI video generation model that topped the Artificial Analysis Text-to-Video leaderboard, surpassing Veo 3, Sora 2 Pro, and other leading competitors. The model demonstrates significant advances in physical accuracy with realistic momentum, fluid motion, and material coherence across frames. Gen-4.5 excels in handling complex physics, fluid dynamics, and human movement, producing cinematic and photorealistic outputs that approach real-world footage quality. While the company acknowledges ongoing challenges with object permanence, the model represents a major leap forward in video generation capabilities. The release is being rolled out to all Runway users over the coming days.

🇨🇳 DeepSeek V3.2 rivals GPT-5 at 30x lower cost 🇨🇳
Chinese startup releases open-source reasoning models under MIT license.

Chinese AI startup DeepSeek launched two powerful reasoning models, V3.2 and V3.2-Speciale, that compete with frontier models while drastically reducing costs and maintaining open-source accessibility under an MIT license. The standard V3.2 matches performance of GPT-5, Claude 4.5 Sonnet, and Gemini 3 Pro across math, coding, and tool-use benchmarks. The more computationally intensive Speciale variant achieved gold medals at prestigious competitions including IMO, IOI, and ICPC 2025, ranking 10th overall at IOI. At just $0.28 input and $0.42 output per million tokens, V3.2 costs nearly 30 times less than Gemini 3.0 Pro. Both 685-billion-parameter models are available on Hugging Face with full weights released, representing a significant advancement in accessible, cost-effective AI capabilities.

INTERESTING TO KNOW

🎥 Kuaishou unveils Kling O1 video platform 🎥

Chinese startup Kuaishou launched Kling O1, a unified AI video system combining generation and editing capabilities in a single model. The platform enables users to generate video clips, swap characters, and perform granular edits within one integrated workflow, representing a new approach to all-in-one video creation tools.

📱 Apple AI chief steps down 📱

Apple's AI chief John Giannandrea has stepped down following repeated delays on an updated Siri assistant. CEO Tim Cook reportedly "lost confidence" in Giannandrea's leadership after the company struggled to keep pace with ChatGPT and other AI assistants. The departure comes as Apple faces mounting pressure to accelerate its AI capabilities.

📩 Have questions or feedback? Just reply to this email , we’d love to hear from you!

🔗 Stay connected: