Gemini 3.1 Flash Lite Arrives

Welcome to another edition of AI PlanetX.

Google launches Gemini 3.1 Flash Lite for speed; OpenAI accidentally reveals GPT-5.4 with 2M-token context; Cursor surges to $2B run rate with enterprise dominance.

• Hottest AI News

• Top AI & SaaS Tools

• AI Tutorial: Build Interactive Simulations

• Top AI & Tech News

• AI Art Spotlight

• Prompt of the Day: Prompt for Cognitive Training

• Featured AI Video

• Course of the Day: Document AI - OCR to Agentic Extraction

Hottest AI News

DeepMind

Google Unveils Gemini 3.1 Flash Lite for Speed and Efficiency

Google has officially launched Gemini 3.1 Flash-Lite, their quickest and most cost-effective model in the Gemini 3 series, built to handle high-volume intelligent workloads at scale without sacrificing quality.

Details:

• It delivers ultra-low latency with 2.5x quicker time-to-first-token and 45% higher output speed than Gemini 2.5 Flash, priced at just $0.25 per 1M input and $1.50 per 1M output tokens

• The model beats larger predecessors with a 1432 Elo on Arena AI and strong scores on GPQA Diamond and MMMU Pro, while adding adjustable “thinking levels” to control reasoning depth

• Available in preview on Google AI Studio and Vertex AI, it’s built for responsive, cost-efficient apps like real-time translation, large-scale moderation, and dynamic UI gen

This release gives developers a quick, smart, and cost‑effective option—see the Google blog for full technical details.

Attio is the AI CRM for modern teams.

Connect your email and calendar and Attio instantly builds your CRM. Every contact, every company, every conversation — organized in one place. Then ask it anything. No more digging, no more data entry. Just answers.

Start your free trial →

OpenAI

OpenAI Accidentally Reveals GPT‑5.4 with 2 Million Token Context

OpenAI may have just accidentally revealed its next major leap, with "GPT-5.4" appearing in multiple code repositories and drop-down menus before being hurriedly scrubbed by developers.

Details:

• Traces of the unreleased model were spotted in GitHub pull requests alongside a "/Fast" command, where explicit references to "gpt-5.4" were quickly renamed to "gpt-5.3-codex" after going viral on X

• The code reveals two major upgrades: a 2M token context window for ultra-long memory, and a “view_image_original_resolution” mode that bypasses compression to analyze UI prototypes and schematics at pixel-level precision

• While ChatGPT 5.2 claims this version, skeptics say it may be hype to rival DeepSeek V4, since a huge context window only matters if recall accuracy holds up

It’s unclear whether this points to an imminent release or a strategic preview, but the signs suggest OpenAI is gearing up for a major leap in memory and visual capabilities.

Top AI & SaaS Tools

• Bytes (Life-time Deal): AI platform that turns documents or prompts into polished, branded micro‑courses you can publish and monetize—no design or technical skills needed

• OpenClaw: Latest update adds Telegram live streaming, ACP subagents by default, a native PDF tool, "openclaw config validate", a pure‑JS Zalo rebuild

• GHOSTYPE: Context-aware macOS voice interface that adapts to the active app, automates workflows, adjusts tone and behavior, and learns user style locally [F-R-E-E]

• getviktor: AI coworker in Slack that connects to 3,000+ tools and proactively runs tasks and automations across your tech stack [F-R-E-E]

• Krisp Accent: Real‑time accent neutralization to American English on the listener’s device, with near‑zero latency and support for Zoom, Teams, and Meet

Speak your prompts. Get better outputs.

The best AI outputs come from detailed prompts. But typing long, context-rich prompts is slow - so most people don't bother.

Wispr Flow turns your voice into clean, ready-to-paste text. Speak naturally into ChatGPT, Claude, Cursor, or any AI tool and get polished output without editing. Describe edge cases, explain context, walk through your thinking - all at the speed you talk.

Millions of people use Flow to give AI tools 10x more context in half the time. 89% of messages sent with zero edits.

Works system-wide on Mac, Windows, iPhone, and now Android (free and unlimited on Android during launch).

Try Wispr Flow free →

AI Tutorial

How to Use Gemini 3 to Build Interactive Simulations in Minutes

If you can describe it, Gemini 3 can probably simulate it. Whether you want to model physics, visualize data, or build small interactive tools, you can do it from a single prompt.

1. Launch Gemini 3

   Go to the Gemini homepage, sign in, and the Gemini 3 chat interface will appear so you can work in the latest environment.

2. Pick right model

   From the model selector, choose what fits your task:

   • Gemini 3 for most reasoning and interactive projects

   • DeepThink for heavy logic and complex simulations (confirm availability in your region)

3. Decide what type of simulation

   Before prompting, quickly define:

   • What system you want to simulate

   • What the user can change (inputs, sliders, buttons)

   • What they should see (graphs, visuals, states, results)

4. Write clear prompt

   In the chat box, describe the goal, controls, and visuals.

   For example:

   "Build a 3D simulation of a solar system with realistic orbital mechanics"

5. Run and inspect output

   Preview or run the generated simulation, try all controls, and quickly check if it behaves as intended and is easy to use.

6. Use Explain to understand

   Tap on Explain or ask follow ups like:

   • "Summarize how the main logic works in simple terms."

   • "Show me where the user inputs change the variables."

   Use this to learn the structure and request small tweaks.

Note: Clearly spell out inputs, outputs, target audience, and platform in your prompt so Gemini 3 can generate cleaner, more useful results.

Top AI & Tech News

• Cursor hit a $2B annualized run rate in February; ~60% from enterprises, with a $29.3B valuation reported in November

• Anthropic’s $60 billion valuation from 200+ investors is threatened by a Pentagon contract dispute that labels the company a supply-chain risk

• Meta is rolling out an AI shopping tool to testers in the U.S. to compete with similar shopping features in ChatGPT and Google’s Gemini

• OpenAI and the Pentagon added safeguards banning intentional tracking of U.S. persons, including through commercial data

AI Art Spotlight

Create Your Own Photo Like This Using the Prompt Below:

Visualize INPUT: [PRODUCT/BRAND]

Create a surreal minimalist photoreal print-ad using the INPUT only. Auto-detect the brand from the INPUT and add the brand’s REAL official logo (not a generic wordmark), high fidelity, correct proportions, clean vector/print look, placed top-left with subtle ink-like texture. Use one bold visual metaphor, 1 hero + 1 accent max, huge negative space, studio lighting, 8K ultra-sharp, premium print-ad composition. Add a two-word slogan bottom-right that matches the brand tone. No clutter, no watermark, no extra text. Aspect ratio 1:1

Model: Nano Banana 2

Prompt of the Day

Personal Prompt for Cognitive Training

This prompt builds a personalized 30–45 minute daily mental training routine by asking 15 sequential questions to map your thinking, surface strengths and weaknesses, and reveal habits that slow you down. Using your answers, it applies evidence based cognitive methods to produce targeted daily exercises, progress tracking, and clear goals for long term improvement.

Featured AI Video

How to Give Your OpenClaw Superpowers (Mission Control)

AI Course of the Day

Document AI: From OCR to Agentic Doc Extraction

This course shows how to build document processing systems that go beyond OCR to accurately extract structured data from PDFs and images. You’ll use LandingAI’s Agentic Document Extraction framework to preserve layout, tables, and context, then deploy the workflow as a serverless RAG application on AWS.