OpenAI has an AI that can think in images?!

Welcome to another edition of Synapze AI Newsletter

This week, we have:

● Breaking news of the week.

● Tutorial of the week - Make your own customGPT.

● AI tools for you.

● Prompt of the day.

NEWS FOR THE WEEK

ChatGPT can now think and visualise in images?!

​OpenAI has launched o3 and o4-mini, its most advanced reasoning models yet. These models are capable of using all ChatGPT tools simultaneously: web browsing, code interpretation, image analysis, file reading and image generation within a single session.

• o3 excels at complex tasks including coding, visual analysis, maths, science, and deep research. It makes 20% fewer major errors than o1 and sets new records on benchmarks like Codeforces and SWE-bench.

• o4-mini is a smaller, faster, and more cost-efficient model that still performs at near-SOTA levels, making it ideal for high-throughput tasks.

• Both models are Agentic they autonomously decide when and how to use tools to solve multi-step problems efficiently.

• They also "think with images", meaning they can interpret, manipulate, and reason over visuals in a more human-like way.

• Safety has been upgraded with a rebuilt training set, new refusal strategies, and high performance on internal safety benchmarks. READ MORE

Why it matters and how can we adapt to this change:

• Tool Use Is Now Native and Intelligent: o3 and o4-mini don’t just access tools, they reason about how and when to use them. This marks a major leap towards AI agents that can handle full workflows without needing prompts for every step.

• Massive Leap in Accuracy and Breadth: With top scores across a wide range of academic and real-world benchmarks, these models set new standards for logic, coding, data analysis, and scientific reasoning.

• Multimodal Mastery: For the first time, models can seamlessly interpret and reason with images, graphs, and screenshots, not just text. This unlocks applications in design, engineering, research, and education.

• Smarter + Cheaper = Scalable AI: o4-mini offers exceptional reasoning performance at lower cost, enabling wider adoption in businesses, education, and development environments without sacrificing quality.

• A True Step Towards Autonomous AI: The models behave more like AI agents, independently navigating tools and datasets to solve problems. This brings us closer to AI systems that execute tasks on your behalf, not just assist with suggestions.

Google Launches Veo 2 Video Creation in Gemini and Whisk

Gemini Advanced now lets users generate 8-second high-resolution videos (720p, MP4) from text prompts. Whisk Animate, part of Google Labs, allows users to animate both text- and image-based prompts into video form using the same Veo 2 model. The model produces realistic, cinematic results with detailed visuals, smooth motion, and lifelike scenes. It’s designed for creative storytelling and visual experimentation. Users can easily share videos on platforms like TikTok and YouTube Shorts. READ MORE 

TUTORIAL FOR THE WEEK

Want to master Prompt Engineering? Here’s how (Prompt Mastery Micro-Course: Normally valued at $97, you can pay what you want for this) → Click the link to view here

In this tutorial. you can create client proposal and presentations within minutes using AI. Check it out HERE: ACCESS THE TRAINING

Check out the latest tutorial on how to use this AI tool to save you hours on YouTube Research HERE: ACCESS THE FULL TUTORIAL

1️⃣ Sign in to NotebookLM
=> Go to https://notebooklm.google
=> Log in using your Google account.

2️⃣ Create a New Notebook
=> Click ‘+ New notebook’ to start a fresh workspace for your research or project.

3️⃣ Add Sources
=> Upload your documents, PDFs, YouTube or links to your Google Docs/slides
Example: Add a marketing strategy PDF or a research paper.
=> You can add as many links on your chosen topic of research to help you learn.

4️⃣ Ask Questions About Your Sources
=> Type natural language questions in the chat panel.
Example: “Summarise the key marketing insights from this PDF.”
=> Create mind maps to help you become an expert with your favourite topics

5️⃣ Highlight Key Quotes and Responses
=> NotebookLM can surface key facts, citations, and passages that match your query. You can save these directly into your notebook.

6️⃣ Give you audio summary or a podcast on your topics
=> You can get NotebookLM to summarise your links in a podcast format…crazy…

7️⃣ Organise Notes and Citations
=> Use the Notebook panel to structure your research findings.
=> Each answer you save becomes part of a citation-backed note.

7️⃣ Export or Share
=> You can copy or export your notes for reports, presentations, or collaboration.

AI TOOLS FOR YOU

Check out my video on my Youtube channel regarding making $$ with faceless videos: Click here for the link

Access the ultimate AI toolbox with over 700+ AI Tools to help you streamline and grow your business: Click here for the link

• Wan 2.1 FLF2V - Alibaba’s open-source First and last frame video model.

• Tavus - The OS for human AI interaction. Building blocks to empower your AI Agent to see, hear, respond, and look human – in real time.

• Make.com - ​A no-code automation platform that enables users to connect apps and design workflows through a visual, drag-and-drop interface.

• OpenAI.fm - ​A free AI voice platform by OpenAI, introduced on March 20, 2025. It focuses on advanced text-to-speech (TTS) and speech-to-text (STT) capabilities.

• Syllaby.io -​ Create Themed AI Faceless Videos with ease. Monetise and create channels on all social platforms and content within minutes.

CHATGPT PROMPT (ChatGPT is your personal content machine - these 7 prompts generate weeks of posts in minutes!)

1️⃣ Content Pillar Architect

Prompt:
“You are a niche content strategist. Based on the niche “[insert niche],” identify 5 evergreen content pillars that align with what my ideal audience wants to learn, solve, or be inspired by. Present each with a short explanation and example topic.”

2️⃣ Scroll-Stopping Hook Generator

Prompt:
“Act as a viral copywriter. Generate 10 bold, curiosity-driven hooks for posts on “[insert topic].” They should be punchy, emotionally charged, and ideal for [insert platform, e.g., Threads or Instagram].”

3️⃣ High-Engagement Post Blueprint

Prompt:
“Turn this idea: “[insert idea]” into a compelling post outline for [insert platform]. Include: a hooky intro, 2–3 key value points, and a clear CTA tailored to the platform’s vibe.”

4️⃣ Omni-Format Repurposing Assistant

Prompt:
“Take this original content: “[insert content]” and repurpose it into:
A tweet thread (7–10 tweets)
A short-form video script (under 60 sec)
A carousel post outline (5–7 slides)”

5️⃣ Magnetic CTA Generator

Prompt:
“Write 10 distinct calls-to-action I can rotate across my posts to boost engagement—focus on encouraging saves, replies, shares, and follows. Tone: [insert brand tone, e.g., playful, authoritative].”

6️⃣ 7-Day Content Scheduler

Prompt:
“Based on my niche “[insert],” create a one-week content plan. Include:
Daily topic ideas
Suggested content format
Goal for each post (e.g., educate, convert, entertain)”

7️⃣ Message Polisher + Tone Matcher

Prompt:
“Here’s a draft post: “[insert text].”
Edit it for clarity, conciseness, and flow. Match the tone to “[insert brand voice or target audience],” and suggest any punchier rewrites if helpful.”