Google Labs has made a groundbreaking announcement, unveiling a major upgrade to its Gemini Pro AI tool. This midsize AI model, which powers the free version of Google’s chatbot, now has the ability to process up to 1 million tokens in preview. This upgrade sets a new benchmark for computational linguistics and machine learning among large language models (LLMs), leaving current leading tools and their 128K capacity in the dust.
According to Google, this upgrade makes Gemini Pro v1.5 theoretically 700% more powerful than OpenAI’s paid GPT-4 model. It boasts “the longest context window of any large-scale foundation model.” Previously, the largest context window available for a publicly accessible large language model was 200,000 tokens. However, Google has significantly increased this capacity, allowing Gemini Pro to consistently run up to 1 million tokens.
While this feature is currently only available for testing purposes, the stable version of Gemini Pro will handle up to 128K tokens. This is a substantial improvement over the 32,000 tokens that Gemini 1.0 can process. Users will have to wait to see the full potential of 1 million tokens.
This upgrade is part of Google’s ongoing efforts to dominate the AI industry. Last week, Gemini Advanced emerged as a credible competitor to ChatGPT Plus. Unlike Anthropic’s Claude, Google’s chatbot is multimodal and provides good results in various tests. It also offers features that OpenAI doesn’t have. However, Gemini Advanced will soon catch up to GPT-4.5 Turbo, which already handles 128,000 tokens.
Gemini Pro 1.5 has demonstrated its versatility through several impressive demonstrations. Google claims that it can process vast amounts of information in one go, including 1 hour of video, 11 hours of audio, codebases with over 30,000 lines of code, or over 700,000 words. The Google Labs team even successfully tested up to 10 million tokens in their research.
One limitation of Gemini models is their inability to analyze PDF files, as pointed out by Decrypt in a comparison between Gemini and ChatGPT. However, Google’s use of Mixture of Experts technology sets Gemini 1.5 apart from its previous versions. This technology, also utilized by Mistral AI, routes requests to a group of smaller “expert” neural networks, resulting in faster and higher-quality responses.
Gemini 1.5 Pro has shown superior performance in several benchmarks compared to Gemini Ultra 1.0, indicating a promising future for Google’s LLMs. Google CEO Sundar Pichai expressed his excitement about the dramatic improvements and comparable quality achieved by Gemini 1.5 Pro while using less compute power.
The announcement did not provide a specific timeline for the release of Gemini Advanced 1.5. However, Google’s enhanced token-processing capabilities will undoubtedly strengthen its position in the AI arms race. As OpenAI actively develops GPT-5, Google’s Gemini Pro is poised to make significant strides in the field of AI.
In conclusion, Google Labs’ unveiling of the Gemini Pro AI upgrade marks a significant milestone in the development of large language models. With its ability to process up to 1 million tokens, Gemini Pro v1.5 surpasses current leading tools and sets a new standard for computational linguistics and machine learning. This upgrade, coupled with the use of Mixture of Experts technology, showcases Google’s commitment to advancing AI capabilities. As the AI industry continues to evolve, Google’s Gemini Pro is poised to make a lasting impact.