Google DeepMind
Google AI research behind Gemini and scientific systems such as AlphaFold.
- Gemini models for multimodal applications
- Research in science, robotics, and reinforcement learning
- Developer access through Google AI and Google Cloud
Google DeepMind combines the former DeepMind and Google Brain teams. Its research includes general-purpose models and scientific systems such as AlphaFold. Gemini is accessed through Google's application, developer, and cloud products rather than a separate DeepMind account.
Gemini and Google AI Studio
Google AI Studio provides a browser interface for trying Gemini and creating a developer API key. The Gemini Developer API connects models to an application's own code, including supported text, image, and multimodal tasks. Model support, quotas, and available tools depend on the selected endpoint.
When to use Google Cloud
Google Cloud AI offers managed cloud model access and custom deployment within a Cloud project. It suits applications that need cloud identity, resource management, regional deployment, or control over model hosting. The developer API and cloud platform have separate account setup and billing arrangements.
How pricing works
Gemini API pricing separates model input, output, caching, and selected tools. Free and paid tiers have different limits and data-use terms. Google Cloud deployments have their own billing and regional options.
Sources
From the archive
- Gemini 3.8 Flash launches alongside a restricted Cyber variant 2026-09-02
- Gemini 2.5 Pro adds reasoning to a long-context model 2025-03-25
- Gemini 2.0 Flash launches as an experimental model 2024-12-11
- SEA-LION v3 builds regional language support on Gemma 2 2024-11-12
- Google releases Gemma 2B and 7B model weights 2024-02-21
- Gemini 1.5 Pro opens a long-context preview 2024-02-15