San Francisco-based OpenAI has launched GPT-4.5, its most advanced artificial intelligence model yet, boasting a significant reduction in hallucinations. The company revealed that the model generates inaccurate information 37% of the time, compared to nearly 60% in its predecessor, GPT-4o.
The release comes amid fierce competition in the AI industry. OpenAI’s move follows launches from rivals, including Anthropic’s Claude 3.7 Sonnet and Elon Musk’s xAI Grok 3. Despite the rise of smaller, more cost-effective AI models like DeepSeek’s R1, OpenAI continues to invest in large-scale language models requiring massive computing power.
OpenAI stated that GPT-4.5 has “broader knowledge and a deeper understanding of the world,” making it more reliable. However, due to high operational costs, the company may limit long-term developer access to the model.
Meanwhile, OpenAI is engaged in major fundraising efforts, reportedly seeking up to $40 billion at a $300 billion valuation. Rival Anthropic is also raising $3.5 billion at a $60 billion valuation.
Microsoft Warns Trump on AI Chip Export Rules
Microsoft has urged the Trump administration to reconsider proposed AI chip export controls, warning they could push American allies toward Chinese technology. Company president Brad Smith cautioned that the regulations, set to take effect in May, would harm U.S. AI leadership and strengthen China’s position.
The proposed rules would restrict AI chip exports to over 100 countries, affecting key U.S. partners like Israel, India, and Singapore. Smith argued that the restrictions would benefit China’s AI sector, particularly companies like Huawei, which is advancing its Ascend 910C processors.
The move has drawn criticism from the European Union, Nvidia, and other tech giants. Microsoft, which plans to spend $80 billion on infrastructure this year, is closely monitoring the policy’s impact on the industry. Nvidia has also engaged with the Trump administration but remains uncertain about the final regulatory outcome.
