Introduction of GPT-4.5

OpenAI has started rolling out its latest AI model, GPT-4.5, to users on its ChatGPT Plus tier. The rollout is expected to take 1-3 days, and the company anticipates changes in rate limits as it learns more about demand.

GPT-4.5 was initially launched for subscribers to OpenAI's $200-a-month ChatGPT Pro plan last week. The model is the largest AI model yet, trained using more computing power and data than any of the company's previous releases.

Capabilities and Limitations

While GPT-4.5 is not necessarily OpenAI's best model, it has shown improvements in certain areas. The model has a deeper world knowledge and higher emotional intelligence, according to OpenAI. It also hallucinates less frequently than most models, which means it is less likely to make things up.

However, GPT-4.5 is very expensive to run, and OpenAI is evaluating whether to continue serving the model in its API in the long term. The company is charging $75 per million tokens (~750,000 words) fed into the model and $150 per million tokens generated by the model, which is significantly more expensive than its workhorse GPT-4o model.

Comparison to Other Models

GPT-4.5 falls short of newer AI 'reasoning' models from Chinese AI company DeepSeek, Anthropic, and OpenAI itself on several AI benchmarks. Despite this, the model has shown impressive capabilities, including convincing another AI to give it cash and tell it a secret code word in one of OpenAI's internal benchmarks.

NEWS FACTORY

Artificial Intelligence

Technology