Gemini 3 this month ✨, OpenAI CFO comments 💰, Snapchat + Perplexity 🤝

Google's Gemini 3 Pro preview model has appeared in VertexAI code, suggesting a November launch. Google has not issued any official statements ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌  ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ ‌ 

Sign Up |Advertise|View Online TLDR Together With TLDR AI 2025-11-06 Move faster with AI: Build secure, maintainable code you can trust (Sponsor) AI is speeding things up, but all that new code creates a bottleneck — who's verifying the quality and security? Don't let unverified code stall your merges. Sonar's automated review gives you the control back, building the trust you need in every line of code, human- or AI-written. With SonarQube, your team can: Use AI without fear: Get continuous analysis for quality and security. Fix issues early: Detect and apply automated fixes before you merge. Maintain your standards: Ensure all code meets your team's quality bar, every time, for long-term code health. Learn how Sonar fuels AI-enabled development while building trust into every line of code. 🚀 Headlines & Launches Google plans to release Gemini 3 Pro Preview in November (1 minute read) Google's Gemini 3 Pro preview model has appeared in VertexAI code, labeled '11-2025', suggesting a November launch. The timeline matches prior testing activity observed in AI Studio. Google has not issued any official statements about the launch. OpenAI Isn't Yet Working Toward an IPO, CFO Says (4 minute read) OpenAI's Chief Financial Officer, Sarah Friar, said the company is not planning to IPO in the short term. It will instead prioritize growth and R&D over profitability. OpenAI is hoping the government will support its offers to increase data center capacity by helping to guarantee the financing for the chips behind its deals. The depreciation rates of AI chips are uncertain, making it more expensive for companies to raise the debt needed to buy them. Friar claims OpenAI could reach profitability on very healthy gross margins if it weren't seeking to invest so aggressively. Google Gemini's Deep Research can look into your emails, drive, and chats (2 minute read) Google Gemini's "Deep Research" now uses Gmail, Drive, and Chat data for AI-generated reports. The feature can create detailed research plans and reports, integrating personal workspace data with web searches. Currently available on desktop, Google plans to roll out mobile access soon. 🧠 Deep Dives & Analysis Semantic Search for Cursor's Coding Agent (4 minute read) Cursor's coding agent improved performance by incorporating custom semantic search over codebases, leading to higher accuracy, fewer user iterations, and better retention of code changes. Thoughts by a non-economist on AI and economics (20 minute read) METR data shows AI task completion capabilities doubling every 6 months, meaning if today's models handle 2-hour tasks, they'll handle 8-hour tasks in a year and full workdays in 18 months. Under more modest projections, where 9% of tasks are automated each year, industrial revolution-scale gains will come within the next two decades. 🧑‍💻 Engineering & Research You don't need 10 different AI tools to deploy an agent (Sponsor) It only takes 1 platform to fast track agentic AI. OutSystems Agent Workbench provides no-code tools, security guardrails, and automated quality reviews so you can deliver human-AI collaboration without the sprawl. Bosch, Heineken, and EDP are already using it to build agentic AI systems. Build with Agent Workbench Enabling Trillion-Parameter Models on AWS EFA (12 minute read) Perplexity published code enabling trillion-parameter mixture-of-experts models to run efficiently across multiple AWS servers for the first time, opening up less advanced data centers to AI inference. The challenge is that large models, even many open weights models, are too large to fit on single servers, but AWS' networking hardware was too slow for the constant inter-server communication these models require. Structured Output in Gemini API (2 minute read) Google enhanced the Gemini API with expanded JSON Schema support and improved property ordering, making structured outputs more reliable for use in multi-agent workflows and database tasks. 🎁 Miscellaneous Pinterest CEO touts open source AI: 'tremendous performance' with reduced costs (4 minute read) Pinterest plans to move forward with open source models for its various use cases. They come at a fraction of the cost of the larger model providers, and they deliver value to users effectively. The company will be observing user feedback to see if people actually want AI shopping agents before implementing them. The company is planning to roll out personalized boards curated with AI. Snap and Perplexity sign $400 million deal to put AI search directly in Snapchat (2 minute read) Perplexity, an AI search engine, will appear directly in Snapchat sometime in early 2026. It will become a prominent part of Snapchat's 'chat' interface. Perplexity is paying Snap $400 million for the integration. Snap may pursue similar partnerships with other AI companies. ⚡ Quick Links The Backend You Don't Have to Touch (Sponsor) 100K+ developers build APIs, secure data, and scale instantly, without touching a single server or config file. See why Xano is the no-code backend that isn't a black box. Explore OpenAI calls for $1T investment to secure US AI leadership (4 minute read) The US risks losing its AI leadership due to energy shortages, threatening economic growth and technological advances. How Two Silicon Valley A-Listers Became Billionaires By Remaking Customer Service With AI (20 minute read) Bret Taylor and Clay Bavor started Sierra, a $10 billion startup that builds AI customer service agents, in 2023. A Short Lesson in Simpler Prompts (5 minute read) You can build much shorter, more effective prompts by working with AI models instead of fighting them and balancing AI with plain old engineering. Love TLDR? Tell your friends and get rewards! Share your referral link below with friends to get free TLDR swag! https://refer.tldr.tech/0b6a6dc1/2 Track your referrals here. Want to advertise in TLDR? 📰 If your company is interested in reaching an audience of AI professionals and decision makers, you may want to advertise with us. Want to work at TLDR? 💼 Apply here or send a friend's resume to jobs@tldr.tech and get $1k if we hire them! If you have any comments or feedback, just respond to this email! Thanks for reading, Andrew Tan, Ali Aminian, & Jacob Turner Manage your subscriptions to our other newsletters on tech, startups, and programming. Or if TLDR AI isn't for you, please unsubscribe.