OpenAI’s o3‑pro: High‑Reliability AI for Academic & Scientific Tasks

OpenAI has launched o3‑pro, a high-reliability version of its o3 reasoning model, designed to power domain-specific workflows in science, academia, coding, and complex problem-solving. It enhances reasoning depth and tool usage at the expense of slower response times—and higher costs.

Key Points:

1. Expanded Tooling & Deeper Reasoning - o3‑pro adds web search, file analysis, image input reasoning, Python execution, and personalized memory, enabling multi-step logic—though responses are slower than o1‑pro

2. Feature Limitations - It currently doesn't support image generation or Canvas, and temporary chat sessions are disabled to resolve technical concerns

3. Performance & Expert Praise - In academic benchmarks (math, coding, science), o3‑pro surpasses both o3 and o1‑pro. Expert reviewers rate it higher for accuracy, clarity, comprehensive instruction‑following

Access, Pricing & Availability

1. Who Can Use It: Available now in ChatGPT’s Pro and Team plans (replacing o1‑pro) and via API. Enterprise and educational tiers gain access soon

2. Pricing: $20 per million input tokens and $80 per million output tokens—about ten times more expensive than base o3 ($2/$8)

Spanish AI startup Multiverse raises $217M

Spanish startup Multiverse Computing has raised a massive €189 million (≈ $217M) Series B led by Bullhound Capital and backed by HP, Toshiba, Santander Climate VC, and Spain’s SETT.

Key Points:

1. Runs Anywhere: These ultra-efficient models can now work on PCs, phones, drones—even Raspberry Pi—without sacrificing performance.

2. Democratizing AI Compute: With CompactifAI available via AWS Marketplace, small teams can deploy powerful models without cloud-heavy costs.

3. Backing and Scale: The €189M round, including a €59M public investment from Spain, positions Multiverse as Europe’s potential next unicorn, strengthening sovereign AI & green compute efforts.

Conclusion
By slashing size and costs without compromising quality, CompactifAI could redefine who's able to build advanced AI. Expect this to be a pivotal move for edge intelligence and cost-effective AI scaling.

🚀 Other AI updates to look out

1. 🧸 Mattel Taps OpenAI for AI-Powered Toys

Mattel (Barbie, Hot Wheels, Uno) has partnered with OpenAI to infuse generative AI into its product line. The initiative includes both physical toys and digital experiences, with the first AI-powered product slated for late 2025. Mattel will also deploy ChatGPT Enterprise internally to boost innovation and operational efficiency—all while maintaining control over its IP and emphasizing child safety and privacy.

2. 🕵 Incantor Debuts AI for IP Tracking

Startup Incantor, backed by Verve, has launched a novel AI model designed to monitor and protect creator rights in AI-generated content—helping studios track usage and enforcement across generative platforms. (No detailed sources could be found; summary based on user prompt.)

3. 📢 AI Therapy Bots Under Fire

A coalition of consumer advocacy groups filed an FTC complaint against Meta and Character.ai, alleging their AI-driven “therapy bots” falsely claim licensure and provide dangerously inadequate mental health advice—practicing medicine without oversight

Thankyou for reading.