Google has officially launched Gemini, its most advanced AI model to date, marking a significant milestone in the company’s AI journey and intensifying competition with OpenAI’s GPT-4.
Three Sizes for Different Needs
Gemini comes in three distinct sizes:
- Gemini Ultra - The largest and most capable model, designed for highly complex tasks
- Gemini Pro - A balanced model for scaling across a wide range of applications
- Gemini Nano - An efficient model optimized for on-device tasks on mobile phones
Multimodal from the Ground Up
Unlike previous AI models that were primarily text-based with added capabilities, Gemini was designed as a natively multimodal system from the start. This means it can seamlessly understand and reason across:
- Text
- Code
- Audio
- Images
- Video
Google claims this native multimodal approach gives Gemini a significant advantage in understanding complex information and solving difficult problems.
Benchmark Performance
According to Google’s internal testing, Gemini Ultra achieves state-of-the-art performance on 30 of 32 widely-used academic benchmarks. Most notably, it scored 90.0% on the MMLU (Massive Multitask Language Understanding) benchmark, becoming the first model to outperform human experts on this test.
The model also excels at coding tasks, scoring 74.4% on the HumanEval benchmark and 67.7% on the Natural2Code benchmark.
Integration Across Google Products
Gemini Pro is already being integrated into Bard, Google’s conversational AI assistant, available in over 170 countries. The company plans to bring Gemini to:
- Google Search
- Google Ads
- Chrome
- Duet AI
- Android devices
Pixel 8 Pro becomes the first smartphone to run Gemini Nano, enabling on-device AI features like summarization in the Recorder app.
Developer Access
Google is making Gemini Pro available to developers and enterprise customers through Google AI Studio and Google Cloud Vertex AI starting December 13, 2023. Gemini Ultra will follow in early 2024 after additional safety testing.
Industry Implications
The launch of Gemini represents Google’s most serious challenge to OpenAI’s dominance in the AI space. With its multimodal capabilities and integration across Google’s vast ecosystem, Gemini could reshape how billions of people interact with AI technology daily.
As CEO Sundar Pichai stated, “This is the beginning of the Gemini era,” signaling Google’s commitment to making this technology central to its future products and services.