OpenAI Launches o3-pro, Its Strongest Reasoning Model Yet

✨TodayOnAI’s Daily Drop

• OpenAI Launches o3-pro, Its Strongest Reasoning Model Yet

• Apple Upgrades AI Models, But Benchmarks Show They Still Lag Rivals

• Google’s AI Overviews Are Gutting News Site Traffic

• 💬 Let’s Fix This Prompt

• 🧰 Today’s AI Toolbox Pick

OPENAI

OpenAI Launches o3-pro, Its Strongest Reasoning Model Yet

🚀 TodayOnAI Insight: OpenAI has launched o3-pro, its most capable AI model to date, replacing o1-pro in ChatGPT Pro and Team. Designed for advanced reasoning, o3-pro significantly outperforms its predecessor—and even top rivals—across critical domains like math, science, and programming.

🔍 Key Takeaways:

• Advanced Reasoning Engine: o3-pro is a stepwise reasoning model optimized for complex tasks in STEM, writing, and business logic.

• Performance Gains: Internal tests show o3-pro outperforms both o3 and competitors like Gemini 2.5 Pro (AIME 2024) and Claude 4 Opus (GPQA Diamond).

• Platform Rollout: Available now to ChatGPT Pro and Team users, with Enterprise and Edu access following next week. API access is already live.

• Pricing: $20 per million input tokens, $80 per million output tokens—consistent with OpenAI’s premium pricing tiers.

• Capabilities & Limits: Supports tools like web browsing, Python, and memory personalization—but lacks image generation, Canvas integration, and currently disables temporary chats due to technical issues.

💡 Why This Stands Out: O3-pro signals a shift toward models explicitly designed for reasoning rather than pattern prediction. Its strong benchmark scores and broader tool integration position it as a serious challenger to multimodal rivals like Gemini and Claude. The tradeoff in speed reflects a deeper bet: that users will prioritize quality and clarity in high-stakes tasks over instant responses. Will this reasoning-first approach become the new standard?

APPLE

Apple Upgrades AI Models, But Benchmarks Show They Still Lag Rivals

🚀 TodayOnAI Insight: Apple has unveiled updates to its Apple Intelligence AI models—but internal benchmarks show they still lag behind key rivals like OpenAI, Google, and Meta. Despite improvements, Apple’s models remain a step behind the cutting edge.

🔍 Key Takeaways:

• Apple released two upgraded models: Apple On-Device (runs locally on iPhones) and Apple Server (runs in Apple data centers).

• Human raters found Apple Server inferior to OpenAI’s year-old GPT-4o and Apple On-Device on par with similar models from Google and Alibaba.

• In image analysis tasks, Meta’s Llama 4 Scout outperformed Apple Server—even though Llama 4 Scout trails leading models from Google and Anthropic.

• Apple On-Device, a ~3B parameter model, now supports summarization, text analysis, and is accessible to developers via the Foundation Models framework.

• Both models now handle ~15 languages and benefit from a broader training dataset spanning PDFs, charts, and infographics.

💡 Why This Stands Out: Apple’s AI push highlights both progress and persistent gaps. While its models now integrate more tightly into iOS and macOS, performance still lags behind industry leaders—a rare admission from Apple. As the company positions itself in the AI race, the question remains: can privacy-focused, on-device AI ever match cloud-based giants?

Google

Google’s AI Overviews Are Gutting News Site Traffic

🚀 TodayOnAI Insight: AI-powered search tools like Google’s AI Overviews are rapidly eroding traffic to news publishers, raising urgent concerns about the sustainability of online journalism in an era where clicks are vanishing.

🔍 Key Takeaways:

• Google’s AI Overviews summarizes search queries without requiring users to click through, sharply reducing referral traffic to news, travel, and product review sites.

• Newer tools like Google’s ChatGPT-style AI Mode present answers conversationally with fewer external links, deepening the impact.

• The New York Times saw a drop in organic search traffic share from 44% in April 2022 to 36.5% in April 2025, per Similarweb data.

• Google claims AI Overviews increases search engagement overall, though not necessarily benefiting publishers.

• Publishers like The Atlantic and The Washington Post are seeking new revenue models, including licensing deals with Amazon, OpenAI, and Perplexity.

💡 Why This Stands Out: The shift from link-based search to AI-generated answers poses a structural threat to the economics of digital journalism. As generative AI reshapes how users consume information, publishers are being forced to rethink business models once built on traffic and ad revenue. Will AI companies step up with fairer content deals, or will journalism be left behind in the algorithm?

✨ See how a simple prompt upgrade can unlock better AI output.

🔹 The Original Prompt

At first glance, this prompt might seem okay. But it's too broad — and that limits the quality of AI-generated results. Let’s improve it using prompt engineering best practices.

✅ The Improved Prompt

💡 Why It's Better

• Clarifies the type of medical company (B2B, tech-focused)

• Defines the target audience (healthcare providers)

• Focuses on valuable content formats (how-tos, case studies, Q&As)

• Aligns topics with business goals (thought leadership + education)

🛠️ Learn how to adapt this prompt for SaaS, AI tools, dev teams & more →
Read the full PromptPilot breakdown

💡 Bonus Tool: Want to generate and master prompts instantly?
👉 Try PromptPilot by TodayOnAI (Free to use)

📰 More from the AI World

• Mistral releases a pair of AI reasoning models

• Google’s AI search features are killing traffic to publishers

• Apple’s new Workout Buddy helps you sweat smarter

• Anthropic’s AI-generated blog dies an early death

🧰 Today’s AI Toolbox Pick

• 🐣TweetHunter (Twitter Tool): An all-in-one Twitter growth tool that constructs viral content.

• 📊Code to Flow (Coding Tool): Visualizes, analyzes, and understands your code through interactive flowcharts.

• ⚙️Bardeen (Automation Tool): A no-code automation platform that produces desired workflows.