2 min read

GPT-5.1 Debuts in Perplexity as China’s Quantum Chip Threatens Nvidia’s AI Dominance

AI Hardware & Quantum Computing

Chinese Optical Quantum Chip Claims 1,000x Speed Over Nvidia GPUs: A Chinese firm has developed an optical quantum chip allegedly 1,000x faster than Nvidia GPUs for AI workloads, with production scaling to 12,000 wafers/year—though yield rates remain low. This could disrupt AI hardware acceleration if validated.


AI Models & Research

GPT-5.1 Released for Perplexity Pro/Max Subscribers: Perplexity AI integrated OpenAI’s GPT-5.1 and GPT-5.1 Thinking into its Pro and Max tiers, alongside upgrades like faster library search and real-time flight data. The model’s broader availability remains unconfirmed.

Gemini 3.0 Release Imminent?: Sundar Pichai’s cryptic social media post and Polymarket predictions (69% chance by Nov 22) suggest Google’s Gemini 3.0 may launch soon, though no official confirmation exists.

Drummer’s Precog Models (24B & 123B): New Precog-24B-v1 and Precog-123B-v1 models generate draft summaries before full responses, improving coherence in storytelling and role-playing. Hosted on Hugging Face.


AI Tools & Developer Innovations

Code-Mode Library: AI Agents as Code Executors: The Code-Mode library lets LLMs write and execute code in a sandbox instead of making tool calls, reducing tokens, API trips, and context overhead. Backed by research from Apple, Cloudflare, and Anthropic.


AI Products & Features

ChatGPT Group Chats Pilot (APAC): OpenAI is testing group chats in ChatGPT (up to 20 users) in Japan, New Zealand, South Korea, and Taiwan, enabling collaborative prompting with user profiles.

Perplexity AI Updates (Nov 14):

  • Comet Assistant: Faster, more accurate, with privacy snapshots.
  • Link Previews: Open links without losing thread context.
  • GPT-5.1 Integration: For Pro/Max subscribers.
  • Real-Time Data: Flight statuses and improved library search.
  • Details in Changelog

Industry & Regulation

Anthropic Advocates for Open-Source AI Regulation: Anthropic warns of AI cyberattacks with minimal human oversight, pushing for open-source model restrictions. Critics like Yann LeCun argue this is a tactic to stifle competition.

Yann LeCun Leaves Meta for "World Models" Startup: The AI pioneer departs to launch a venture focused on "world models", amid broader industry shifts like AI data center hardware shortages and OpenAI’s $7B+ annual inference costs.