OpenAI has introduced GPT-4.5, its latest general-purpose large language model, aimed at improving accuracy and user experience. The company stated that GPT-4.5 exhibits fewer inaccuracies compared to its predecessors, GPT-4o and o1. Initially, access will be granted to software developers and ChatGPT Pro subscribers, who pay $200 per month.
Enhanced Capabilities and Access Tiers
According to OpenAI, early tests reveal that GPT-4.5 delivers a more natural interaction experience. It boasts a broader knowledge base, better comprehension of user intent, and improved emotional intelligence (EQ). These enhancements make it particularly useful for refining writing, programming, and tackling practical problems.
GPT-4.5 is expected to roll out to ChatGPT Plus and Team subscribers next week, with ChatGPT Edu and Enterprise users gaining access the following week, according to OpenAI’s research lead, Alex Paino.
CEO’s Perspective and Market Impact
OpenAI CEO Sam Altman acknowledged that while GPT-4.5 may not top industry benchmarks, it offers a unique “magic” he has not seen before. This release is a key milestone for OpenAI, which is reportedly in talks to raise $40 billion in funding, valuing the company at a staggering $340 billion.
The launch also comes amid increasing competition, particularly from Chinese AI lab DeepSeek. In January, concerns over DeepSeek’s advancements contributed to Nvidia’s market cap plummeting by nearly $600 billion in a single day.
Challenges and Future Developments
Altman noted that GPT-4.5 is a “giant, expensive model” and that OpenAI is facing constraints due to a shortage of GPUs. The company plans to deploy “tens of thousands” of additional GPUs in the coming weeks to support the rollout of GPT-4.5 to its broader subscriber base.
Microsoft, a major OpenAI investor, announced that GPT-4.5 will be available for preview through Azure AI Foundry. Microsoft has invested over $13 billion into OpenAI and plays a critical role in supplying computing resources for its models.
Looking Ahead
Altman has hinted that GPT-4.5 will be OpenAI’s last “non-chain-of-thought” model before transitioning to more advanced reasoning-based AI systems. Meanwhile, the company is preparing for a broader upgrade with the upcoming GPT-5, which is expected to incorporate an even wider range of OpenAI’s technologies.
