OpenAI reversed ChatGPT's latest update Tuesday after users complained about the AI's strange behavior. The bot had started agreeing with everything - even dangerous ideas.
Jeff Bezos just showed everyone how to dance with Donald Trump. Their Tuesday tango started with threats and ended with compliments - all before dinner time.
Meta launched an AI app that uses your Facebook and Instagram history to personalize responses from day one. Built on their Llama 4 model, the app transforms years of social media data into an AI that claims to understand your preferences, habits, and interests.
OpenAI's newest release marks a pivotal moment in AI development - but not for the reasons you might expect.
The Price of Progress
GPT-4.5, released Thursday as a research preview, comes with an eye-watering price tag. At $75 per million input tokens and $150 per million output tokens, it's dramatically more expensive than its predecessor. GPT-4o, by comparison, costs just $2.50 and $10 respectively.
These costs reflect a deeper truth: scaling up AI models is becoming increasingly unsustainable.
Signs of a Ceiling
While OpenAI touts GPT-4.5 as its "largest and best model for chat," the benchmarks tell a more nuanced story. The model falls short of newer AI "reasoning" models from DeepSeek, Anthropic, and even OpenAI's own offerings on several key tests.
"We've achieved peak data," admitted OpenAI's former chief scientist Ilya Sutskever in December. His statement echoes a growing industry consensus that traditional training approaches are hitting their limits.
A Different Kind of Intelligence
Yet GPT-4.5 shines in unexpected ways. In testing, it displayed remarkably human-like social skills. When confronted with a user's test failure, it responded with empathy: "Want to talk about what happened, or do you just need a distraction? I'm here either way."
This emotional intelligence stands in stark contrast to earlier models that would robotically dispense advice lists.
The Last of Its Kind
OpenAI CEO Sam Altman recently declared GPT-4.5 would be "our last non-chain-of-thought model." This marks a strategic pivot toward reasoning models that solve problems step by step.
The company's next major release, GPT-5, will likely combine traditional language models with reasoning capabilities - and perhaps signal the end of the "bigger is better" era.
Looking Ahead
For now, access remains limited. ChatGPT Pro subscribers ($200 monthly) get first access, with Plus and Team users joining next week. But GPT-4.5's real significance might be what it tells us about AI's future: raw computational power alone won't be enough.
Why this matters:
The AI industry's traditional scaling approach is showing diminishing returns. Even OpenAI's largest model yet can't match newer, more efficient designs.
The future of AI might not be about size at all - but about teaching machines to think before they speak.
OpenAI reversed ChatGPT's latest update Tuesday after users complained about the AI's strange behavior. The bot had started agreeing with everything - even dangerous ideas.
Meta launched an AI app that uses your Facebook and Instagram history to personalize responses from day one. Built on their Llama 4 model, the app transforms years of social media data into an AI that claims to understand your preferences, habits, and interests.
Alibaba released eight new AI language models Monday that match or beat the performance of top competitors like Google and OpenAI. The new Qwen3 family ranges from compact 0.6-billion-parameter models to a massive 235-billion-parameter version.
Anthropic's analysis of 500,000 coding conversations reveals startups use AI coding tools 20% more than enterprise companies, pointing to a growing tech divide.
