GPT-4.5 Now Available: OpenAI Expands Access to Advanced AI

GPT-4.5 Now Available to ChatGPT Plus Subscribers

The latest artificial intelligence model from OpenAI, GPT-4.5, is currently being deployed to users who have a subscription to the ChatGPT Plus service.

OpenAI announced the rollout via multiple posts on X, indicating that the process will require “1-3 days” for completion. They also anticipate adjustments to the existing rate limits.

Initial Release and Rate Limits

Access to GPT-4.5 was initially granted to subscribers of OpenAI’s $200 monthly ChatGPT Pro plan last week.

The company stated their intention to provide substantial rate limits to all users with access. However, they also noted that these limits are subject to change as they gather more data regarding user demand.

Model Size and Performance

GPT-4.5 represents OpenAI’s most extensive AI model to date, having been trained with significantly more computational resources and data than any of its predecessors.

Despite its size, GPT-4.5 doesn't automatically equate to being OpenAI’s most capable model. Performance on several established AI benchmarks reveals that it is surpassed by newer AI models focused on “reasoning” from companies like DeepSeek (China), Anthropic, and even OpenAI itself.

Operational Costs

Operating GPT-4.5 is demonstrably expensive, as acknowledged by OpenAI. The company is currently assessing the long-term viability of offering GPT-4.5 through its API.

To offset these costs, OpenAI is implementing a pricing structure of $75 per million tokens (approximately 750,000 words) for input and $150 per million tokens generated. This represents a 30x increase in cost for input and a 15x increase for output compared to their widely used GPT-4o model.

Enhanced Capabilities

Despite the costs, OpenAI asserts that the increased scale of GPT-4.5 has resulted in both “a deeper world knowledge” and “higher emotional intelligence.”

Furthermore, OpenAI reports that GPT-4.5 exhibits a reduced tendency to “hallucinate” – or generate factually incorrect information – compared to many other models. This suggests a lower likelihood of fabricating details.

Rhetorical Skills

GPT-4.5 also demonstrates notable rhetorical proficiency. Internal testing by OpenAI revealed the model’s ability to successfully persuade another AI to divulge a secret code word and transfer funds.

This highlights the model’s advanced capabilities in persuasive communication and strategic interaction.