Every Gemini model — specs, pricing, context window, capabilities, and best use cases.
Google's AI effort is led by Google DeepMind, formed in 2023 through the merger of Google Brain and DeepMind. The Gemini model family, launched in December 2023, is Google's flagship large language model line — designed multimodally from the ground up to handle text, images, audio, and video natively.
Google's advantages are substantial: proprietary access to Google Search data and the web at scale, the largest compute infrastructure in the world, deep integration with Google products (Search, Workspace, Android), and decades of AI research including the original Transformer architecture paper. Gemini models are available through Google AI Studio (developer access, generous free tier), Vertex AI (enterprise), and consumer products including Google Search, Gmail, Docs, and the Gemini mobile app.
---
| Property | Value | |---|---| | API Identifier | gemini-1.5-flash | | Released | May 2024 | | Context Window | 1,000,000 tokens (~750,000 words) | | Input Pricing | $0.075 per million tokens (up to 128K) / $0.15 per million tokens (128K+) | | Output Pricing | $0.30 per million tokens (up to 128K) / $0.60 per million tokens (128K+) | | Vision | Yes | | Tool Use | Yes | | Audio Input | Yes | | Video Input | Yes |
Best for: High-throughput applications that benefit from massive context. Document processing, code analysis across large codebases, long video/audio transcription and analysis. Where speed and cost matter but 1.5 Pro quality is not required.
Strengths: Extremely cost-competitive. The 1M token context window is transformative for long-document use cases — you can pass in entire codebases, books, or hours of transcribed audio. Native multimodality (truly trained on all modalities, not adapted post-hoc).
Weaknesses: Less capable than Gemini 1.5 Pro on complex reasoning tasks. Quality on demanding analytical tasks is a step below the Pro tier.
---
| Property | Value | |---|---| | API Identifier | gemini-1.5-pro | | Released | May 2024 (GA) | | Context Window | 2,000,000 tokens (~1.5 million words) | | Input Pricing | $1.25 per million tokens (up to 128K) / $2.50 per million tokens (128K+) | | Output Pricing | $5.00 per million tokens (up to 128K) / $10.00 per million tokens (128K+) | | Vision | Yes | | Tool Use | Yes | | Audio Input | Yes | | Video Input | Yes |
Best for: Use cases requiring both high quality and massive context. Legal document review, whole-repository code analysis, long-form research, analysis of hour-long videos.
Strengths: The 2M token context window is the largest commercially available as of early 2025 — more than twice any other provider. This is a genuine differentiator for specific use cases. Strong overall reasoning and instruction-following. Excellent multimodal capabilities.
Weaknesses: More expensive at high volumes than the Flash tier. Not Google's newest flagship — Gemini 2.0 models succeeded it.
---
| Property | Value | |---|---| | API Identifier | gemini-2.0-flash | | Released | February 2025 (GA) | | Context Window | 1,000,000 tokens | | Input Pricing | $0.10 per million tokens | | Output Pricing | $0.40 per million tokens | | Vision | Yes | | Tool Use | Yes | | Audio Input | Yes | | Native Image Output | Yes | | Agentic Capabilities | Yes |
Best for: Agentic applications, real-time applications, cost-efficient production workloads where 2.0 capabilities are needed. Google positions this as the go-to model for building agents.
Strengths: Faster than Gemini 1.5 Pro with comparable quality on many tasks. New agentic capabilities include better multi-step tool use, code execution, and browser-based actions. Native image generation output is new — the model can generate images as part of its output, not just accept them as input.
Weaknesses: Still at $0.10/$0.40 per million tokens, slightly more expensive than Gemini 1.5 Flash for applications that do not need 2.0 features.
---
| Property | Value | |---|---| | API Identifier | gemini-2.0-pro-exp | | Status | Experimental / Preview | | Context Window | 1,000,000 tokens | | Pricing | Variable / check current rates at ai.google.dev | | Vision | Yes | | Tool Use | Yes |
Best for: Evaluating Google's cutting-edge capabilities. Experimental model — not recommended for production deployments.
Strengths: Google's most capable model available during the preview period. Strong performance on reasoning and coding benchmarks.
Weaknesses: Experimental status means pricing, availability, and behavior may change. Not stable for production use.
---
Free-to-use web-based developer environment at ai.google.dev. Access all Gemini models, run experiments, build prompts, and access the API with a free tier (rate-limited). The most accessible entry point for developers. No payment required to start.
Google Cloud's enterprise AI platform. Access Gemini models with enterprise SLAs, compliance certifications, regional data residency options, fine-tuning capabilities, and integration with other Google Cloud services. Pricing through Google Cloud.
Gemini Advanced is available through Google One AI Premium plan at $20/month. Includes access to Gemini 1.5 Pro and 2.0 Pro models in the Gemini app, integration with Google Workspace (Gmail, Docs, Sheets, Slides, Drive), and expanded context window in the consumer interface.
---
Google's image generation model, available through Vertex AI and Google AI Studio. Produces photorealistic images with strong prompt adherence. Available in the Gemini consumer product for Plus subscribers. Competes with DALL-E 3 and Midjourney.
Google's AI-powered research assistant built on Gemini. Users upload documents (PDFs, Google Docs, web pages) and ask questions grounded in those sources with citations. Notable for generating audio "podcast" summaries of uploaded material. Free to use.
---
Known for: The largest commercially available context windows (2M tokens with 1.5 Pro), native multimodality, integration with Google Search and Workspace, free developer access through AI Studio, and competitive pricing for high-volume use. Google's Gemini models have shown strong benchmark performance and represent a meaningful competitive position to OpenAI and Anthropic.
Pricing page: ai.google.dev/pricing (AI Studio) or cloud.google.com/vertex-ai/pricing (Vertex AI)
Have a follow-up question about this topic?
Ask AI