Helping to share the web since 1996

Gemini 1.5 : Google’s latest AI

Introducing Gemini 1.5, Google’s latest AI model featuring significant enhancements over its predecessor. Google continues its series of AI releases, following the recent rebranding of Google Bard to Gemini, the launch of the Gemini Advanced app, and the introduction of a premium AI plan.

Just unveiled on Thursday, Gemini 1.5 surpasses its predecessor, Gemini 1.0, which was released in December. The upgraded model boasts a more extended context window, improved comprehension, and overall superior performance.

Google CEO Sundar Pichai highlighted the remarkable capabilities of Gemini 1.5 Pro, the initial release for early testing. Pichai noted that this version achieves comparable quality to the previously announced 1.0 Ultra, Google’s most advanced large language model, while utilizing less computational resources. Pichai emphasized the potential of longer context windows, enabling developers to build more useful models and applications.

To achieve its enhanced performance, Gemini 1.5 adopts a new version of the Mixture-of-Experts (MoE) architecture. This architecture allows the model to learn and selectively activate the most relevant pathways in its neural network, thereby improving efficiency.

Google asserts that Gemini 1.5 Pro can process up to one million tokens in production, a significant increase from the 32,000 tokens processed by Gemini 1.0. The model’s ability to handle more tokens is crucial, as it directly impacts the depth of information it can absorb, resulting in more informed responses.

According to Google, 1.5 Pro can handle extensive data in one session, such as processing up to one hour of video, 11 hours of audio, and codebases with over 30,000 lines of code or more than 700,000 words.

In a demonstration, Google showcased 1.5 Pro’s capabilities by providing it with a 44-minute silent Buster Keaton movie. The model swiftly processed the content and answered a variety of questions, including multimodal queries.

Gemini 1.5 Pro exhibited impressive performance against benchmarks, surpassing 87% of the benchmarks used by Google for LLM development. Additionally, it excelled in the Needle In A Haystack (NIAH) evaluation and Machine Translation from One Book (MTOB) benchmarks, assessing the model’s acuity and learning abilities.

To address user concerns about safety, Google emphasizes extensive evaluations to ensure the secure and responsible deployment of Gemini 1.5 Pro.



Back to news headlines