MODELS
Gemini 3.5 Flash
by Google DeepMind
Overview
Fast Gemini 3.5 model for reasoning, coding, long-context, multimodal, and agentic tool-use workflows.
Details
Gemini 3.5 Flash is a Google DeepMind Gemini model described in official sources as a fast model for reasoning, coding, long-context, multimodal, and agentic tool-use workflows. The Gemini API model page lists model code gemini-3.5-flash, supports text, image, video, audio, and PDF inputs with text output, and documents a 1,048,576-token input limit and 65,536-token output limit. Official docs also list capabilities such as function calling, structured outputs, grounding, thinking, URL context, code execution, and computer use preview.
When to Use
Use for Gemini API applications that need a fast Gemini 3.5 model with reasoning and coding support. Use for workflows requiring long-context input with the documented 1 048 576-token input limit. Use for multimodal input workflows involving text image video audio or PDFs with text output. Use when an application needs supported Gemini API capabilities such as function calling structured outputs grounding URL context code execution or computer use preview.
Getting Started
- Review the official Gemini 3.5 Flash model page and model card from Google DeepMind.
- Use the Gemini API model code gemini-3.5-flash from the official model documentation.
- Check the Gemini API pricing page for current paid rates before production use.
- Review Gemini API safety settings documentation
- including configurable harm-block thresholds for gemini-3.5-flash.
Key Features
- •Fast Gemini 3.5 model for reasoning
- •coding
- •long-context
- •multimodal
- •and agentic tool-use workflows.
- •Model code: gemini-3.5-flash.
- •Supports text
- •image
- •video
- •audio
- •and PDF inputs with text output.
- •Documents a 1
- •048
- •576-token input limit and 65
- •536-token output limit.
- •Supports function calling
- •structured outputs
- •grounding
- •thinking
- •URL context
- •code execution
- •and computer use preview.
Capabilities
- •reasoning
- •coding
- •long-context
- •multimodal input
- •text output
- •function calling
- •structured outputs
- •grounding
- •thinking
- •URL context
- •code execution
- •computer use preview
- •configurable safety settings
Last updated Jul 16, 2026