In partnership with

Welcome back!

Gemini’s new “Nano Banana” image model is sparking unexpected viral applications and workflows. What seemed just like another model upgrade is becoming a showcase of how quickly AI tools evolve beyond their authors’ anticipations. Let’s unpack…

Today’s Summary:

• 🍌 Google’s Nano Banana unlocks new workflows

• 📲 Google Translate adds language practice

• 🗣️ OpenAI launches gpt-realtime model

• 💻 xAI debuts Grok Code Fast 1

• 🎮 Mirage 2 makes images playable

• 🎬 Google Vids brings free AI editing

• 🛠️ 2 new tools

TOP STORY

Google Nano Banana starts a wave of new workflows

The Summary: Shortly after Google’s launch of “Nano Banana” Gemini 2.5 Flash Image editing model, the internet has flooded with new use cases. People are turning sketches into 3D environments, redesigning backyards, labeling screenshots, and restoring family photos. What looked like a routine upgrade is revealing unexpected versatility and novel use cases that even researchers hadn’t anticipated.

Use cases and workflows:

• World-building: Maps become navigable environments

• Object modeling: Turn photos of buildings into isometric views

• Clothing try-on: Test new outfits on your photo

• AR annotations: Annotate photos with educational overlays

• Stage a home: Drop new sofas, paint walls, redesign your backyard

• Family archives: Restore and colorize old photos

• Photo headshots: Clean up or restyle profile photos

• Mix ideas: Combine multiple images into one concept

Why it matters: Nano Banana seems to be a universal image manipulator. It can restore, repair, recolor, mix, and reimagine parts of an image with uncanny accuracy while keeping everything else intact. That turns what used to be a painstaking Photoshop workflow into a plain chat sentence. It unlocks a style of creativity closer to sketching on paper or opening a new medium of expression that anyone can tap into.

Try Google Nano Banana for free in Gemini

FROM OUR PARTNERS

Write 3x Faster Just by Speaking

AI voice dictation that's actually intelligent

Typeless turns your raw, unfiltered voice into beautifully polished writing - in real time.

It works like magic, feels like cheating, and allows your thoughts to flow more freely than ever before.

With Typeless, you become more creative. More inspired. And more in-tune with your own ideas.

Your voice is your strength. Typeless turns it into a superpower.

Download for Mac today

GOOGLE

Google brings live talk and language practice to Google Translate

The Summary: Google Translate is rolling out two AI-powered features: live conversation translation and a personalized language learning mode. Users can now practice speaking and listening with specific scenarios that adapt to skill levels and goals. The update uses Gemini AI for more natural interactions and smoother translations, with early access starting in the US, India, and Mexico.

Key details:

• Google Translate Live conversation mode handles accents, pauses, and background noise for real-time dialogue

• Language practice is launching first for English, Spanish, French, and Portuguese learners

• Both features are built on Google’s Gemini AI models with multimodal reasoning and speech recognition

• Currently available in the US, India, and Mexico for Android and iOS

Why it matters: Google Translate is now doubling as a personal language tutor. Embedding adaptive practice in this free app puts Google in direct competition with Duolingo as the casual learner’s first stop. Because of Google Translate’s massive scale, even modest features may create the world’s largest language learning platform almost instantly.

How-to

FROM OUR PARTNERS

Build Your Own AI Agent in 2 Minutes

The Simplest Way To Create and Launch AI Agents

Imagine if ChatGPT and Zapier had a baby. That's Lindy. Build AI agents in minutes to automate workflows, save time, and grow your business.

Let Lindy's agents handle customer support, data entry, lead enrichment, appointment scheduling, and more while you focus on what matters most - growing your business.

Get 400 Free Credits

OPENAI

OpenAI launches gpt-realtime model with live voice and image support

The Summary: OpenAI has released gpt-realtime and made its Realtime API generally available, introducing advanced voice capabilities, SIP phone integration, and image input support. The new model processes speech directly, reducing latency while sounding more natural and expressive. Early benchmarks show major gains in reasoning, instruction following, and tool usage.

Key details:

• gpt-realtime scored 82.8% on Big Bench Audio (up from 65.6%), 30.5% on MultiChallenge, and 66.5% on ComplexFuncBench

• New voices Cedar and Marin added; existing eight voices upgraded for higher audio quality

• Realtime API now supports remote MCP servers, SIP integration, reusable prompts, and image input alongside text/audio

• Pricing cut by 20%: $32 per million input tokens, $64 per million output tokens.

Why it matters: This release gives developers the tools to build production-ready AI voice apps instead of just demos. It enables AI agents to handle customer calls, guide users through forms, or explain what’s in a screenshot, all in real time. The model has low latency and makes speech more natural. Lower pricing and direct links into phone systems and enterprise apps make it easier for businesses to ship useful, scalable voice products.

QUICK NEWS

Quick news

• Google Vids makes AI editing free with new tools

• xAI launches Grok Code Fast 1 cheap and fast coding model

• Mirage 2 turns images into playable game worlds

TOOLS

🥇 New tools

• Wanderboat 2 - Social, Local, AI map search

• HyNote - Audio transcription, meeting notes, PDF summary