WOWWWW! So much happening atm it’s hard to keep up.
Google released Nano Banana Pro this week. Time to fire your designer, photographer and editor. Here’s a sneak peak at whats possible…
📌 TL;DR
Google's Nano Banana Pro → Best image gen model available. Opens doors for product shots, ad creatives, social content, and automated workflows.
Claude Opus 4.5 → Top coding model and great for short-form copy and complex multi-step tasks. + context limits fixed across all Claude models.
GPT-5.1 Pro → Slow, methodical model for when accuracy matters. Best for deep research, launch plans, market analysis, strategic briefs. Expect 10min + waits for answers.
ChatGPT group chats → Now live globally. Add up to 20 people into shared conversations.
Google's Nano Banana Pro just replaced your designer
Google released Nano Banana Pro this week (image generation powered by Gemini 3 Pro). This is a massive upgrade over the already-amazing Nano Banana, which ran on Gemini 2.5 Flash.
Key upgrades:
Better text representation → Text renders better than ever now. It still struggles a bit with intricate, small text like fine print on packaging, but for headlines, logos, and body copy its pretty damn good.
Reasoning on top of image generation → The model thinks before it generates. Eg; you can upload a report and say ‘make an infographic’ and will distill the info down
High-fidelity image editing → Upload an image, tell it what to change, and it will get it spot on.
Resolution bumps → 1k, 2k, and 4k outputs. Finally.
I've been testing this all week, and it's the best image generation model I've used by a country mile. Nanobanana Pro opens up so many doors for online business owners.
Product photos
Custom icon sets
Social media content
Ad creatives
Professional headshots
Accurate graphs/ infographics
YT thumbnails
Carousel posts
Image editing
Logos/ branding
+ so much more
This also opens doors for automated workflows. Think agents that take your YouTube video and spit out a thumbnail in your style. Or an agent that converts one long-form video into 5 Instagram carousels with custom graphics. Ad creative agents that generate 20 variations in minutes.
Simple, plain-English prompts work best. Don't overthink it. Just describe what you want like you're talking to a human.
When you upload reference images, clearly define the role of each one. For example: "Use Image A for the character's pose, Image B for the art style, and Image C for the background environment."
How to access Nano Banana Pro:
You can access it directly through Gemini. Just make sure you select 'Thinking' mode first, then Tools → Create images. Not recommended though, since you can't control the aspect ratio or resolution.
You can also access it through Google AI Studio, but you'll need to connect your API key.
My personal favourite is to use an aggregator like Higgsfield, OpenArt, or Artlist. You get access to all the best image and video models in one place, with proper control over aspect ratios, resolution, and settings.
Here’s some cool shit I’ve done to get your creative juices flowing…
“Add the hiking pack in image A onto the tarmac floor in the bottom right of the scene in image B”
“Using the reference image: generate a clean white studio product shot of this exact watch fully disassembled into a precise vertical exploded stack, floating with the strap undone. Soft diffused top light, sharp focus, accurate metal and fabric texture, luxury commercial realism.”
“Add the rolex watch from image A into the red Tray in image B. keep the watch true to size. the watch is larger compared to the car porsche keys
“Create a shiny chrome 3D [OBJECT] icon floating freely in mid-air with no floor or surface visible. Bright white studio lighting with clean, crisp reflections on a pure white background.”
“Create a instagram post text graphic exactly the same style as image A, accept the text should say '3 Free AI tools you should be using to grow your business'“
“Using the Golden Goose sneakers from image 1, create a clean [ANGLE] product shot on the same background color as image 1. Soft studio lighting, full shoe in frame.”
“Place the coffee table from C image into the scene in image A. Place the book in image B on the coffee table.”
OpenAI released GPT-5.1 Pro
GPT-5.1 Pro is a slow, heavy-weight reasoning model. When given really tough problems, it feels smarter than anything else I've used.
Matt Shumer says “It feels like a different class of system that has been told: You have plenty of time. Do not screw this up."
Where it excels:
Deep research, launch plans, market analysis, strategic briefs. Anything with lots of constraints or 10-20 decision points instead of 2-3. If you tell it "do X, but don't touch Y, and keep Z the same," it actually does that. You'll use what it gives you.
Where it loses:
Creative writing & Speed. It takes forever. That's the tradeoff. Waiting 10 minutes for an answer is not ideal. For quick tasks, it's not worth the wait.
For most people, other models like Gemini 3 stays the daily driver. For anything that requires deep thought, planning, and research, and anything that I need to get right on the first try, I reach for GPT-5.1 Pro.
OpenAI also rolled out group chats across all ChatGPT plans. You can add up to 20 people into a shared conversation with ChatGPT, where everyone sees the same messages and responses. Your personal memory stays separate from group chats. Not a crazy impactful feature, but could be useful for brainstorming sessions/ collaboration switching tools.
Anthropic released Claude Opus 4.5
Anthropic dropped Claude Opus 4.5 this week. It crossed 80% on SWE-Bench Verified, making it the top coding model available right now.
The Every team tested Opus 4.5 by running 11 projects simultaneously over six hours. None of them derailed. Previous models would lose track of important details or bleed context between tasks. Opus 4.5 held the thread on all of them.
Where it wins for non-technical work:
Short-form writing and promotional copy. Opus 4.5 produced the strongest promotional copy of the three models tested (Sonnet 4.5, GPT-5.1, Opus 4.5). The prose was punchy, specific, and avoided typical AI writing patterns.
Complex multi-step tasks. If you're working on something with lots of moving parts, constraints, or decision points, Opus 4.5 maintains focus better than previous models. It doesn't spiral into overwrought solutions or lose the thread.
Where it struggles:
Editing work. It's too gentle as an editor. When tested against Sonnet 4.5 and GPT-5.1 on the same draft, Opus 4.5 called it "pretty solid" and flagged only 7 issues. Sonnet 4.5 flagged 47 issues and called for a complete structural overhaul. If you need an AI to push back and make your work better, stick with Sonnet 4.5.
Anthropic also fixed the context limit problem this week for all models in the Claude app. Long conversations no longer hit a wall. Claude now compacts earlier messages automatically, so the chat just keeps going.
Anthropic also launched Claude for Excel. Max, Team, and Enterprise users can now use Claude directly in Excel. Opus 4.5 handles complex spreadsheet work much better than earlier models.
💡 Prompt of the week
Give Gemini 3 a few images in a style you want to replicate into new designs and use this prompt to replicate the style for new images.
“Help me write a prompt that can be used to replicate this exact image style to other [illustration/ graphic/ photo] assets. Do not write any specific details about this image, only capture the essence and style of the image so it can be used to make other images in this same consistent style”
🛠 Tools to try
Downie → This thing just lives in my toolbar and saves me 10 minutes every day. Drop any link (YouTube, Instagram, TikTok, whatever) and it downloads the file instantly. No more searching "YT to mp3" and waiting through sketchy sites. Little tools like this save you 5 mins per day… but when you stack enough of them and you become a productivity machine.
Google Mixboard → AI-powered visual canvas from Google Labs. Perfect for moodboarding and planning the style for brands and campaigns.
💌 Earn free gifts
I’m giving away my personal Claude skills library (playbooks for AI).
Just share this newsletter with 2 friends (who are into AI), and it’s all yours! You currently have {{rp_num_referrals}} referrals, only {{rp_num_referrals_until_next_milestone}} away.
or send them this link: {{rp_refer_url}}
Catch you next week,
