🎙 YES! Music generation is now live in the bot. FOR FREE.

Google has released the Lyria 3 music generator, and it’s already available to users of our bot.

✅ Lyria can create melodies in any language, based on a description or just a single image. Realistic vocals and clear, natural lyrics. Cover art for your songs is made by Nano Banana.

❓ How to create a Hit:

1️⃣ Go to @GPT4AgentsBot
2️⃣ Start your request with “Generate a song…”
3️⃣ Describe the melody, specify the genre, and vocal language.
4️⃣ For a track based on an image, attach an atmospheric picture and caption it “Generate a song”
✨30 tracks per week for free!

Try it now ➡️ @GPT4AgentsBot

🟩 GPT, help me!

#GPTprompt: Create viral phone wallpapers ⚡

Level up your lock screen with this killer prompt! Our bot will whip up a stylish family portrait in just a few minutes in a cartoon, designer, or photorealistic style.

✅ Choose a model: Nano Banana 2, Nano Banana Pro or GPT Image 1.5. Each one interprets style differently, so try them out and compare the results.

❓ How to create a photo like this:
1️⃣ Open @GPT4AgentsBot
2️⃣ Tap Menu → /image → choose a model
3️⃣ Upload photos in the correct order and paste the prompt*:

➡️ Realistic style:

Reference 1 — MAN
Reference 2 — WOMAN
Reference 3 — GIRL [6 years old]
Reference 4 — BABY BOY [3 years old]
Create a single photorealistic image combining all [four] people from the reference photos.

IMPORTANT: Preserve the exact facial features, face proportions, hairstyle, hair color, age and identity of each person exactly as in the reference images. Do not change their appearance even slightly.

Scene composition:
All [four] characters are playfully peeking from behind a vertical matte wall on the left side of the frame. Only their heads, hands and upper torsos are visible.
They are arranged in a perfectly straight vertical line, one under another, as if hiding behind the wall and curiously looking out.

Order from top to bottom:
1. Man
2. Woman
3. Girl (6)
4. Baby boy (3)
All characters gently hold the wall with their hands and lean forward slightly with playful, curious expressions.

Clothing:
Man — black T-shirt
Woman — black tank top with thin straps
Girl — black T-shirt
Baby boy — black T-shirt

Environment:
Soft pastel textured white wall in the foreground on the left.
Clean smooth studio gradient background in light tones on the right.

Lighting:
Soft studio lighting, warm gentle highlights, no harsh shadows.
Materials and details: Velvety matte wall texture, natural skin detail, realistic hair strands, soft clothing fabric.

Style: High-end photorealistic studio photography, cozy family mood, shallow depth of field, centered framing.
Ultra-clean render, 4K realism. Format — [9:16].

*Change the text in [square brackets] to suit your family members.

👀 Examples in pictures. More prompts in the comments👇

❤️ — exactly what I wanted
🔥 — thanks, I’ll try it
🐳 — not my thing

🟩 GPT, help me!

#GPTprompt: Saving Family Photo Archive with Banana Pro!

The latest AI models can restore old photos with stunning realism and precision. Cracks, creases, blur, faded details are no longer a problem.

➡️ Bringing old photos back to life:
1️⃣ Open @GPT4AgentsBot
2️⃣ Tap Menu → /image → choose Nano Banana Pro
3️⃣ Upload your photo and paste this prompt:

Upscale and enhance the attached source image to true 8K UHD (7680×4320) with a 4x upscale factor, delivering a masterpiece-level, ultra-detailed, hyper-realistic result. Preserve the exact same face with 100% facial similarity-no alterations, morphing, or distortion of facial features. Maintain original skin and surface textures while subtly enhancing micro-details and increasing sharpness (1.5) for crisp clarity. Apply low denoising strength (0.25) to retain natural realism. Enhance with cinematic lighting, sharp focus, intricate texture depth, seamless high-quality gradients, and vibrant yet natural colors with professional color correction and strong gradient refinement. Avoid any plastic, over-smoothed, blurry, waxy skin, flat textures, pixelation, noise, JPEG artifacts, washed-out tones, CGI/cartoon/3D-render look, loss of detail, distorted eyes, or identity changes. Use DPM++ 2M Karras sampler, 50 steps, and CFG scale 7.0 to ensure a photorealistic, raw photography finish that significantly improves clarity and depth while staying faithful to the original image.

4️⃣ Download the result in 3–5 minutes.

👀 Check out the before/after examples in the images. Give it a try and share the magic with your loved ones.

❤️ — pure magic!
🔥 — definitely trying this

🟩 GPT, help me!