DeepMind
Pioneering AI research lab, known for advancements in deep learning and AI for games.
Last verified:
What is DeepMind?
Google DeepMind is an AI research laboratory that builds next-generation AI systems safely and responsibly to benefit humanity. The lab combines breakthrough research in deep reinforcement learning, machine learning, neuroscience, and engineering to create transformative AI technologies including Gemini models, AlphaFold for protein structure prediction, Veo for video generation, Lyria for music generation, and Genie for world model generation.
Key features include frontier AI models like Gemini 3.5 with agentic coding and advanced multimodal understanding, AlphaFold revealing millions of 3D protein structures, Genie 3 generating interactive environments, Veo 3.1 creating high-quality video with audio, Lyria 3 composing music with vocals, and Gemma open models for developers. The platform offers Google AI Studio for building with AI models, Google Antigravity as an AI-first development platform, Gemini API for integration, and the Gemini Enterprise Agent Platform for scaling agents.
This tool is for AI researchers, data scientists, developers building AI applications, scientists advancing scientific discovery, enterprises needing AI agents for automation, robotics developers, and anyone interested in cutting-edge AI technology. It serves organizations like Shopify, Macquarie Bank, Salesforce, Ramp, Xero, and Databricks for tasks ranging from merchant growth forecasting to automated admin workflows.
The mission is to achieve artificial general intelligence (AGI) while prioritizing safety and responsibility, solving hard scientific challenges, advancing science, transforming work, serving diverse communities, and improving billions of people's lives through breakthrough technologies.
DeepMind pricing
Pricing model: Freemium
Free tier: Free input and output tokens with limited model access, Google AI Studio access, content used to improve products. Paid tier: Higher rate limits, context caching access, Batch API with 50% cost reduction, access to most advanced models, content not used for product improvement. Gemini 3.5 Flash Standard: $1.50 per 1M input tokens, $9.00 per 1M output tokens. Gemini 3.1 Flash-Lite Standard: $0.25 per 1M input tokens (text/image/video), $0.50 (audio), $1.50 per 1M output tokens. Batch pricing: 50% discount on standard rates. Priority pricing: 1.8x standard rates. Enterprise tier: All paid features plus dedicated support, advanced security and compliance, provisioned throughput, volume-based discounts. Veo 3.1: $0.05-$0.60 per second depending on model and resolution. Lyria 3: $0.04 per 30s song, $0.08 per full song. Imagen 4: $0.02-$0.06 per image. Context caching: $0.01-$0.27 per 1M tokens per hour depending on model. Grounding: 5,000 prompts/month free for Gemini 3, then $14 per 1,000 search queries.
DeepMind pros
- Breakthrough AI research including AlphaGo, AlphaFold, and Gemini models
- Gemini 3.5 leads benchmarks in coding, agentic workflows, and multimodal reasoning
- Free tier with generous limits for developers getting started
- AlphaFold reveals millions of protein structures advancing biology research
- Veo 3.1 generates high-quality video with native audio generation
- Lyria 3 composes music with vocals and acoustic details
- Genie 3 creates diverse interactive world environments
- Gemma open models maximize intelligence-per-parameter
- Context caching reduces costs for repeated prompts
- Batch API offers 50% cost reduction for bulk processing
- Advanced agentic capabilities for long horizon task execution
- State-of-the-art robotics with Gemini Robotics vision-language-action model
- Grounding with Google Search and Maps for real-time information
- Enterprise platform with dedicated support and advanced security
- Transformer architecture invention underpinning modern LLMs
- Deep reinforcement learning pioneer with DQN and AlphaZero
- WeatherNext provides state-of-the-art extreme weather forecasts
DeepMind cons
- Free tier content used to improve Google products
- Paid tier required for production applications with higher volumes
- Preview models may change before becoming stable with restrictive rate limits
- Grounding with Google Search costs $14 per 1,000 queries after free tier
- Image generation expensive at $0.045 to $0.24 per image depending on resolution
- Video generation costs $0.05 to $0.60 per second based on quality and model
- Context caching storage costs $0.50 to $1.80 per million tokens per hour
- Some advanced models like Gemini 3.5 Pro not yet available
- Voice naturalness and verbosity still improving in audio models
- Rate limits more restrictive on preview models compared to stable versions
Frequently asked questions about DeepMind
What is Google DeepMind's mission?
Google DeepMind's mission is to build AI responsibly to benefit humanity. The team works to create the next generation of AI systems safely and responsibly, solving hard scientific and engineering challenges to create breakthrough technologies that advance science, transform work, serve diverse communities, and improve billions of people's lives. The ultimate goal is achieving artificial general intelligence (AGI).
What AI models does DeepMind offer?
DeepMind offers Gemini (frontier intelligence with action), Gemini Omni (create anything from video), Gemma (open models), Nano Banana (image generation/editing), Lyria (music generation), Veo (video generation), Imagen (text-to-image), Gemini Audio (real-time audio), Gemini Robotics (vision-language-action), Gemini Embedding (multimodal embeddings), and Genie 3 (world model).
What is the free tier limit for the Gemini API?
The free tier offers free input and output tokens with limited access to certain models and Google AI Studio access. For Gemini 3.5 Flash, input and output are completely free. Free tier includes 5,000 prompts per month for Grounding with Google Search (shared across Gemini 3). Content generated on free tier is used to improve Google products.
What is AlphaFold and what does it do?
AlphaFold is an AI system that accurately predicts 3D models of protein structures, catalyzing a new wave of progress in biology. It reveals millions of intricate 3D protein structures and helps scientists understand how life's molecules interact. This breakthrough has advanced scientific understanding of biological systems.
What is Gemini 3.5 best used for?
Gemini 3.5 Flash is best for frontier performance across agents and coding, tackling complex development tasks with advanced reasoning at speed. It excels at agentic coding, advanced multimodal understanding, executing long horizon tasks, and multi-step problem-solving. It leads benchmarks in Terminal-bench coding (76.2%), SWE-Bench Pro (55.1%), and MCP Atlas workflows (83.6%).
How does context caching work and what does it cost?
Context caching allows you to cache context for repeated prompts to reduce costs. Storage costs range from $0.50 to $1.80 per 1,000,000 tokens per hour depending on the model. Caching prices vary by model: Gemini 3.5 Flash costs $1.00 per million tokens per hour, Gemini 3.1 Flash-Lite costs $0.025-$0.05 depending on input type, and Gemini 3 Pro costs $0.05-$0.18 depending on input type.
What is Google Antigravity?
Google Antigravity is Google's AI-first development platform that allows anyone to be a builder. It provides the fastest path from prompt to production and is one of the main ways to start building with DeepMind's cutting-edge AI models and tools alongside Google AI Studio and Gemini API.
What is Genie 3?
Genie 3 is a general purpose world model that can generate an unprecedented diversity of interactive environments. It represents a new frontier for world models and enables creating worlds with Genie, simulating real-world places with Project Genie and Street View.
How does Veo video generation pricing work?
Veo 3.1 Standard video with audio costs $0.40 per second for 720p and 1080p, $0.60 per second for 4k. Veo 3.1 Fast costs $0.10 per second for 720p, $0.12 for 1080p, $0.30 for 4k. Veo 3.1 Lite costs $0.05 per second for 720p, $0.08 for 1080p (4k not supported). Veo 3 Standard is $0.40 per second, Veo 3 Fast ranges from $0.10 to $0.30 per second.
What enterprise features are available?
The Gemini Enterprise Agent Platform offers all paid tier features plus optional dedicated support channels, advanced security and compliance, provisioned throughput, volume-based discounts based on usage, ML ops capabilities, Model Garden access, and the ability to build, scale, and govern agents for large-scale deployments with custom security and compliance needs.