Google has been touting the capabilities of its flagship generative AI models, Gemini 1.5 Pro and 1.5 Flash, as powerful tools that can process and analyze vast amounts of data. However, recent research suggests that these models may not be as effective as advertised. Studies have shown that Gemini models struggle to correctly answer questions about large datasets, with accuracy rates as low as 40-50%. Despite being able to process long contexts of up to 2 million tokens, these models still fall short when it comes to understanding the content they are analyzing. In fact, when tested on tasks like evaluating true/false statements about fiction books or reasoning over videos, Gemini models performed poorly, often below random chance. This raises concerns about Google's overpromising with these models, as customers seek more reliable and effective generative AI solutions. Researchers advocate for better benchmarks and third-party critique to ensure transparency and accuracy in evaluating the capabilities of AI models.