Google · Gemini 2.5

Gemini 2.5 Flash

Cheap multimodal at million-token scale.

Gemini 2.5 Flash is a multimodal AI model developed by Google, part of the Gemini family, and released in 2023. It is designed to handle diverse input modalities, including text, image, audio, video, and files, making it highly versatile for complex tasks across various domains. What sets Gemini 2.5 Flash apart is its expansive context window of 1,048,576 tokens, allowing for the processing of large-scale data inputs in a single session and enabling deeper understanding and coherent reasoning.

The model combines advancements in transformer architecture with optimized multi-modal integration, making it particularly effective at tasks requiring both extensive context comprehension and alignment across different data types. It achieves a balance between cost-effective input processing and premium output quality, providing strong performance within reasonable computational constraints.

Gemini 2.5 Flash occupies a flagship-tier position in the Gemini lineup, offering significant improvements in context window size and multimodal processing efficiency compared to previous versions. It enhances token management and modality alignment, allowing for seamless processing of large and varied datasets.

Background

Gemini is a family of multimodal large language models (LLMs) developed by Google DeepMind, and the successor to LaMDA and PaLM 2. Comprising Gemini Pro, Gemini Deep Think, Gemini Flash, and Gemini Flash Lite, it was announced on December 6, 2023. It powers the chatbot of the same name.

Wikipedia

Specs

Context window: 1.0M tokens
Max output: 66K tokens
Input ($/1M tokens): $0.300
Output ($/1M tokens): $2.50
Modalities: File · Image · Text · Audio · Video
Weights: Closed

Pricing last synced Apr 27, 2026 via OpenRouter. Confirm against official docs before committing.

Capabilities

Tool use
Vision
Extended thinking
Prompt caching
Open weights

What it excels at

Large Context Window
Processes over 1 million tokens within a single input for nuanced understanding of lengthy documents or datasets.
Multimodal Integration
Handles text, images, audio, video, and files with strong alignment and reasoning capabilities.
Cost-Effective Input Processing
Optimized for cheaper input handling at $0.3/1M tokens.
Scalable Efficiency
Balances high performance with reasonable computational costs, suitable for diverse applications.

When to use this model

→Document Analysis and Summarization — Supports processing of lengthy or complex datasets in one pass using its large context window.
→Content Creation — Generates rich outputs based on multimodal inputs, ideal for creative industries.
→Multimodal Data Analysis — Excels at integrating and analyzing data from text, images, video, and audio simultaneously.
→Educational Tools Development — Facilitates the creation of interactive and comprehensive learning materials from diverse content types.
→Media and Entertainment Production — Enables dynamic multimedia content generation, processing images, videos, and audio seamlessly.

Analysis synthesized from gpt-4o, llama-4-maverick, phi-4, etc.

API model id

gemini-2.5-flash

Vendor docs: ai.google.dev/docs

Compare Gemini 2.5 Flash with