🚀 Gemini 3.0 Launched

Hey folks,

Google just rolled out AI travel canvas this week, and it's the kind of feature that makes you realize we're past the chatbot era. You tell it where you want to go, and it generates a full itinerary, finds flight deals, and lets you book everything without leaving the canvas. No tab-switching between search results, no copying flight numbers into booking sites, just conversational planning that ends with a confirmation email. It's the latest example of agentic AI moving from demos to products people actually use.

Let's dive in..

🤖 Gemini 3.0 Shipped Immediately

🐝 The Buzz: Google launched Gemini 3.0 with same-day deployment across Search, Gemini app, AI Studio, and Vertex AI. Sundar Pichai said "Our most intelligent model that helps you bring any idea to life." It crushes GPT-5.1 on multimodal tasks, 87.6% vs 80.4% Video-MMU and the first model to hit 1501 Elo on LMArena.

• Day-one Search deployment: First time shipping a Gemini model in Search immediately at launch. Unlike previous releases requiring weeks of rollout, Gemini 3 went live instantly for 2 Billion Search users.

• Deep Think for extended reasoning: Deep Think mode enables complex multi-step reasoning by spending more compute during inference, similar to GPT-5.1's reasoning approach but with multimodal capabilities across text, audio, video, images.

• Crushes GPT-5.1 on multimodal and abstract reasoning: Video-MMU 87.6% vs 80.4%, ARC-AGI-2 31.1% vs 17.6% (nearly double), CharXiv Reasoning 81.4% vs 69.5%. The 6x leap over Gemini 2.5 on ARC-AGI-2 (4.9% → 31.1%) signals capability jumps beyond task optimization toward general intelligence benchmarks.

• Google Antigravity: New platform integrating editor, terminal, and browser access for autonomous agents. Combined with Gemini 3's reasoning, enables end-to-end application development, from concept to deployment with AI handling coding, testing, and iteration while developers provide high-level direction.

💡 Takeaway: The day-one Search deployment represents Google's confidence in production readiness with no phased rollout. The strategic bet is multimodal + agentic capabilities justify premium over text-only models. Deep Think mode matters for applications requiring extended reasoning where compute cost trades off against accuracy.

Together with Neurons

Make Every Platform Work for Your Ads

Marketers waste millions on bad creatives.
You don’t have to.

Neurons AI predicts effectiveness in seconds.
Not days. Not weeks.

Test for recall, attention, impact, and more; before a dollar gets spent.

Brands like Google, Facebook, and Coca-Cola already trust it. Neurons clients saw results like +73% CTR, 2x CVR, and +20% brand awareness.

Get a 14-day FREE trial & start improving your ads, at top speed.

🚀 Grok 4.1 Is Emotionally Aware

🐝 The Buzz: xAI launched Grok 4.1 with the highest LMArena score ever recorded, 1483 Elo, 31 points above any non-xAI model. It boasts 65% fewer hallucinations and emotion-aware responses that detect frustration, hesitation, and sarcasm.

• #1 LMArena ranking: Grok 4.1 Thinking holds the top spot, while the non-reasoning mode ranks #2 at 1465 Elo, passing every other model's full-reasoning configuration.

• 65% hallucination reduction: Internal evaluations show error rates dropped from 12% to 4% on real-world information seeking.

• Emotion-aware responses: Ranked top 3 for creative writing, the model detects nuanced emotional signals including frustration, hesitation, sarcasm, excitement and adjusts responses without "scripted AI monologue" patterns.

💡 Takeaway: The 31-point Elo lead signals a meaningful capability gap, but the real story is reliability: 65% fewer hallucinations. Engineering teams should evaluate for production workflows where factual accuracy matters more than creative ideation.

📰 Like newsletters? Here are more our readers enjoy.

💻️ How to Prompt Guide For GPT 5.1

If you’re working with AI models, the new prompt-guide from OpenAI for GPT-5.1 is solid help. Rather than “write better prompts”, it shows how to tweak your style for advanced reasoning, coding, tool-use and agentic tasks.

• Go to OpenAI Cookbook

• Read & Practice

Example Before: “Write a summary of this meeting transcript”

After (Optimised for GPT-5.1):

You are a senior product manager summarising a 60-minute design review meeting. Provide:

A one-paragraph overview of key decisions.
A bullet list of 3 action items with owners and due dates.
Highlight any open questions needing follow-up.

Use reasoning_effort: medium, verbosity: medium.

This upgraded prompt uses explicit structure + task definition + parameter hints (reasoning/verbosity) aligning to the guide.

🐝 AI buzz bits

🏦 Asia's DBS bank expects AI to drive $768 million in revenue this year. CEO Tan Su Shan told CNBC "It's not hope, it's now" after a decade of AI implementation across institutional services and internal analytics.

💰 OpenAI's infrastructure costs hit $8.65 billion for the first nine months of this year, up from $3.8B for all of 2024, while Microsoft received $865.8 million in revenue share.

🖥️ Windsurf and Cursor revealed their proprietary coding models run on Chinese AI systems. Windsurf confirmed its core Composer model is provided by Z.ai, while users discovered similar patterns in Cursor's SWE-1.5.

📊 Meta is baking "AI-driven impact" into performance reviews, making AI usage a core expectation across all roles. They will be judged on how they use AI to deliver results and build productivity tools. AI proficiency becomes a job requirement, not a nice-to-have skill.

🔥 AI productivity tools

For a full list of 1000+ AI tools, visit our Supertool Directory

🤘 CD Burner Wednesday