OpenAI Unveils New Powerful AI Models o3 and o4 Mini

OpenAI has unveiled two powerful new AI models—o3 and o4-mini—designed to tackle complex reasoning tasks in coding, math, and science. The o3 model is OpenAI’s most capable reasoning engine to date, while o4-mini offers a more affordable option with impressive performance. Both models leverage ChatGPT’s full toolkit, including web browsing, code interpretation, and enhanced visual understanding, enabling them to solve multi-step problems with greater independence and accuracy.

In addition to these models, OpenAI introduced Codex CLI, an open-source command-line tool that bridges natural language and code. Developers can now describe desired outcomes and have Codex CLI generate, debug, and understand code directly on their local machines. Supporting all OpenAI models, including the latest GPT-4.1, Codex CLI accelerates prototyping and development workflows, making AI-powered coding more accessible and efficient.

Meanwhile, Google has launched Gemini 2.5 Flash in preview via the Gemini API on Google AI Studio and Vertex AI. This hybrid reasoning model allows developers to set "thinking budgets," optimizing the balance between response quality, speed, and cost, and delivers improved performance over previous versions even when reasoning is minimized.

Let’s explore what these innovations mean for the future of AI-powered enterprises.

OpenAI Launches o3 and o4-mini AI Models, Enhancing Reasoning Capabilities

OpenAI has released o3 and o4-mini, two new AI models designed to tackle complex reasoning problems in coding, math, and science. The o3 model is OpenAI’s top-performing reasoning model to date, while o4-mini is a more affordable version delivering impressive results in similar areas. These models can utilize all of ChatGPT’s tools, including web browsing, code interpretation, image analysis, and image generation, enabling them to solve multi-step problems and act more independently.

The visual abilities of o3 and o4-mini have also been enhanced, allowing users to upload diagrams, notes, or sketches for the models to understand and reason through. In addition to the new AI models, OpenAI introduced Codex CLI, a tool for developers that connects the AI directly with their code on local machines. The newly launched models are available to users on ChatGPT Plus, Pro, and Team plans, with o3-pro, a more advanced version of o3, coming soon for Pro subscribers.

OpenAI Eyes Windsurf Acquisition as It Unveils Codex CLI

OpenAI is in discussions to acquire Windsurf for approximately $3 billion, a strategic move to broaden its capabilities. Simultaneously, OpenAI has introduced Codex CLI, an open-source coding agent that translates natural language into functional code. Codex CLI empowers developers to efficiently build, fix, and understand code by describing their desired outcome and supports all OpenAI models, including the latest GPT-4.1. The tool allows users to create applications and debug codebases directly from their command line, offering rapid prototyping and development capabilities. Codex CLI is available on GitHub.

Google Unveils Gemini 2.5 Flash with Smarter, Faster AI

Gemini 2.5 Flash, an upgrade to 2.0 Flash, is now in preview via the Gemini API in Google AI Studio and Vertex AI. It features enhanced reasoning capabilities, speed, and cost-effectiveness. As a hybrid reasoning model, developers can control the "thinking" process, setting budgets to balance quality, cost, and latency. Even with "thinking off," it maintains the speed of 2.0 Flash while improving performance. The model decides how much to "think" based on task complexity, with a thinking budget ranging from 0 to 24576 tokens. It is available in a dedicated dropdown in the Gemini app.

This new level of control allows developers to tailor AI responses more precisely to their needs, making Gemini 2.5 Flash a versatile tool for a wide range of applications, from simple queries to complex problem-solving tasks.

Hand Picked Video

In this quick tutorial, I show you how to effectively use Olly Social's Auto Commenter feature on LinkedIn.

Top AI Products from this week

• Omakase.ai - Omakase AI is an AI-powered platform that enables e-commerce businesses to instantly create a personalized shopper agent by simply entering their website URL, automating customer service and boosting online sales through hyper-personalized, 24/7 shopping assistance and intelligent recommendations—all without any coding or complex setup required

• Shipable - Shipable.ai helps you transform data into powerful AI Agents. It allows founders, marketers, and support teams to automate customer conversations across web, WhatsApp, and Slack without code. It is designed for simplicity, and engineered for security.

• Shotup AI - Shotup AI is a tool that acts as a photographic memory, saving and learning from your screenshots. By capturing screen images, the AI learns context and visuals, allowing you to quickly retrieve and reference the information later through chat.

• Omniflow - Omniflow is an AI-driven tool designed to streamline product development. It helps in crafting product requirement documents, creating full-stack applications, and automating task creation and resource planning for teams.

• Artefact - Artefact is an AI-powered platform that enhances documents with smart suggestions, validation, and contextual feedback tailored to your team's needs, boosting collaboration and productivity.

• Veo 2 in Gemini - Google Gemini Video Generation is a tool that creates short, high-quality videos from simple text descriptions, making video creation easy and fast.

This week in AI

• Chatgpt Memory with Search - ChatGPT now offers fast, accurate web search with source links to all users across devices, improving access to up-to-date information.

• Hence AI - Hence AI has introduced Hence Global, an AI-powered risk advisor that helps companies monitor and manage geopolitical and trade risks amid rising global tensions, providing daily updates and tailored insights at an affordable price, making expert risk management accessible to businesses of all sizes.

• Grok Chatbot Adds Memory Feature - xAI’s Grok now remembers past chats to offer personalized responses, matching ChatGPT and Gemini’s capabilities. The feature is in beta and gives users control over their data.

• Apple & Google’s AI Smart Glasses Race - Apple and Google are developing AI-powered smart glasses to compete with Meta, focusing on real-time translation and augmented reality features to lead the future of wearable tech.

• Notion MCP Server - Notion MCP Server implements an MCP server for the Notion API, limiting exposed API scope for security. Configure the integration, add MCP config to your client.