Gemini 1.5 Pro

Model Overview

Gemini 1.5 Pro is Google DeepMind's flagship mid-size multimodal model, delivering highly capable performance across a wide range of tasks. Its standout feature is a revolutionary massive context window, allowing it to process and reason over an unprecedented amount of information in a single prompt.

Capabilities

• Massive Context Window: Gemini 1.5 Pro features a standard context window of 1 million tokens, expandable up to 2 million tokens. This allows it to process vast amounts of information simultaneously, such as 2 hours of video, 22 hours of audio, over 60,000 lines of code, or over 1.4 million words.
• Native Multimodality: It is built from the ground up to understand and process text, images, audio, and video natively, without relying on separate models for different modalities.
• In-Context Learning: With its massive context window, Gemini 1.5 Pro excels at learning new skills directly from information provided in the prompt, reducing the need for fine-tuning.
• Advanced Reasoning: It demonstrates significant improvements in complex reasoning tasks, math, and coding compared to its predecessor, Gemini 1.0 Pro.

Example Use Cases

• Large-Scale Data Analysis: Analyzing entire financial reports, legal documents, or entire books in a single query.
• Video Understanding: Analyzing long-form videos to extract specific information, summarize plot points, or answer questions about visual details.
• Codebase Comprehension: Providing the model with an entire software repository to understand its architecture, find bugs, or suggest improvements.

Performance & Benchmarks

Gemini 1.5 Pro achieves state-of-the-art performance across numerous benchmarks, often exceeding the performance of larger models:

• MMLU (General Knowledge): 81.9%
• HumanEval (Code): 84.1%
• MATH: 58.5%
• Needle In A Haystack (Recall): 99% recall accuracy across its 1 million token context window.

About Google DeepMind

Google DeepMind is a world-leading artificial intelligence research laboratory, formed by the merger of Google Brain and DeepMind, dedicated to solving intelligence to advance science and benefit humanity.