Gemini
by Google DeepMind
Google's most capable multimodal AI model for complex reasoning and creative tasks
Visit Product
107 upvotes
3,480 views
About
Gemini is Google's most advanced AI model family, designed to be natively multimodal — meaning it can understand and reason across text, images, audio, video, and code simultaneously from the ground up, rather than being retrofitted with these capabilities. Available through Google's Gemini app, Google Workspace, and the Gemini API, it represents Google's answer to the rise of ChatGPT.
Gemini comes in three sizes: Ultra (the most capable, competing with GPT-4), Pro (for a wide range of tasks), and Nano (for on-device applications). Gemini Ultra scored higher than human experts on MMLU (Massive Multitask Language Understanding) and outperformed GPT-4 on 30 of 32 academic benchmarks at its launch. The model has a 1 million token context window in its 1.5 Pro version — the largest context window of any generally available LLM.
Gemini is deeply integrated into Google's product ecosystem, powering AI features in Gmail, Google Docs, Google Search (AI Overviews), Google Photos, and Android. For developers, Gemini is accessible through Google AI Studio and Vertex AI with competitive pricing.
Gemini comes in three sizes: Ultra (the most capable, competing with GPT-4), Pro (for a wide range of tasks), and Nano (for on-device applications). Gemini Ultra scored higher than human experts on MMLU (Massive Multitask Language Understanding) and outperformed GPT-4 on 30 of 32 academic benchmarks at its launch. The model has a 1 million token context window in its 1.5 Pro version — the largest context window of any generally available LLM.
Gemini is deeply integrated into Google's product ecosystem, powering AI features in Gmail, Google Docs, Google Search (AI Overviews), Google Photos, and Android. For developers, Gemini is accessible through Google AI Studio and Vertex AI with competitive pricing.
Product Features
- Native multimodal: text, images, audio, video, and code
- Gemini 1.5 Pro: 1 million token context window
- Deep integration with Google Workspace (Docs, Gmail, Sheets)
- Google Search integration with AI Overviews
- Gemini Advanced: access to Gemini Ultra capabilities
- Code generation and debugging across multiple languages
- Real-time information via Google Search grounding
- Gemini API for developers with generous free tier
- On-device Gemini Nano in Android for privacy-first AI
- Available in 230+ countries and territories
- Gemini 1.5 Pro: 1 million token context window
- Deep integration with Google Workspace (Docs, Gmail, Sheets)
- Google Search integration with AI Overviews
- Gemini Advanced: access to Gemini Ultra capabilities
- Code generation and debugging across multiple languages
- Real-time information via Google Search grounding
- Gemini API for developers with generous free tier
- On-device Gemini Nano in Android for privacy-first AI
- Available in 230+ countries and territories
About the Publisher
Gemini is developed by Google DeepMind, formed by the merger of Google Brain and DeepMind in April 2023. Google DeepMind is led by Demis Hassabis and is one of the world's foremost AI research organizations, responsible for AlphaGo, AlphaFold, AlphaStar, and numerous other landmark AI achievements. Google has invested tens of billions of dollars in AI research and infrastructure and employs thousands of AI researchers and engineers worldwide.