> ## Documentation Index
> Fetch the complete documentation index at: https://docs.zerotwo.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Google (Gemini)
> Gemini models in ZeroTwo — up to 1M token context, built-in reasoning, and multimodal capabilities.
Google DeepMind's Gemini models bring exceptionally large context windows, built-in reasoning capabilities, and strong multimodal performance to ZeroTwo. Gemini 2.5 and 3.1 models are among the most capable models available.
## Available models
| Model | Context | Strengths | Premium? |
| ------------------------------ | ------- | ------------------------------------------------------- | -------- |
| **Gemini 3.1 Pro (preview)** | 1M | Cutting-edge; Google's latest flagship | Yes |
| **Gemini 3.1 Flash (preview)** | 1M | Fast and efficient at 1M context | Pro+ |
| **Gemini 2.5 Pro** | 1M | Strong reasoning, excellent overall quality, multimodal | Yes |
| **Gemini 2.5 Flash** | 1M | Fast, efficient; very capable for the speed | Pro+ |
| **Gemini 2.5 Flash Lite** | 1M | Ultra-efficient; standard tier | Standard |
| **Gemini 2.0 Flash** | 128k | Fast, reliable; proven model | Standard |
| **Gemini 1.5 Pro** | 1M | Previous generation flagship; still highly capable | Pro+ |
## Gemini's strengths
### Massive context windows (up to 1M tokens)
Gemini's most distinctive feature is its context window. Most models top out at 128k or 200k tokens. Gemini 2.5 Pro and above offer **1 million tokens** of context — enough to fit:
* A 750,000-word novel
* An entire software codebase with thousands of files
* Many hours of meeting transcripts
* Hundreds of research papers
This makes Gemini uniquely suited for tasks involving very large amounts of text.
### Built-in reasoning (Gemini 2.5+)
Gemini 2.5 Pro and newer models include built-in reasoning capabilities. These models think through complex problems before responding, producing significantly better results on math, logic, coding challenges, and structured analysis — without requiring a separate reasoning model.
### Multimodal capabilities
Gemini models are natively multimodal — they can process text, images, and (for some models) video in a single conversation. Send images alongside your prompts for visual analysis, diagram interpretation, or image-to-text tasks.
### Google integration potential
Gemini models are built with awareness of Google's ecosystem, making them well-suited for tasks involving web-grounded information (when web tools are enabled) and Google Workspace-style workflows.
## Model details
### Gemini 3.1 Pro (preview)
Google's latest and most capable model. State-of-the-art performance across reasoning, coding, writing, and multimodal tasks. Available as a preview in ZeroTwo — subject to early access limitations.
**Best for**: The most demanding tasks where you want Google's latest capabilities.
### Gemini 3.1 Flash (preview)
The fast variant of Gemini 3.1 — maintains the 1M context window and strong capability but optimized for speed and efficiency. Preview availability.
**Best for**: Fast tasks that benefit from large context or 3.1's latest capabilities.
### Gemini 2.5 Pro
Google's proven flagship. 1M context, built-in reasoning, excellent coding, writing, and analysis. Gemini 2.5 Pro is one of ZeroTwo's top-recommended models for quality-critical work.
**Best for**: Long document analysis, complex coding, research synthesis, multimodal tasks. Strong general-purpose model.
### Gemini 2.5 Flash
The fast, efficient version of Gemini 2.5. Maintains the 1M context window and strong capability but generates faster and costs less of your quota. An excellent balance of quality and efficiency.
**Best for**: Most everyday tasks at scale — especially those involving large documents but where maximum reasoning depth isn't needed.
### Gemini 2.5 Flash Lite
Ultra-efficient standard-tier model. Available to all plans including Free. Fast and capable for routine tasks.
**Best for**: Quick tasks, high-frequency use, and Free-plan users who want a capable, fast model.
### Gemini 2.0 Flash
Reliable, fast standard-tier model with 128k context. A proven workhorse for general-purpose tasks.
**Best for**: Reliable everyday use, standard tasks, Free-plan users.
### Gemini 1.5 Pro
Previous generation flagship with 1M context. Still highly capable — a solid choice when 2.5 Pro is quota-constrained.
## Choosing a Gemini model
| Use case | Recommended model |
| ----------------------------------- | ----------------------------------------- |
| Very long documents (> 200k tokens) | Gemini 2.5 Pro or 3.1 Pro |
| Quality-critical work | Gemini 2.5 Pro or 3.1 Pro |
| Fast tasks, large context | Gemini 2.5 Flash |
| Everyday use, Free plan | Gemini 2.5 Flash Lite or Gemini 2.0 Flash |
| Latest Google capabilities | Gemini 3.1 Pro (preview) |
| Multimodal (image + text) | Gemini 2.5 Pro |
Gemini 2.5 Flash is an excellent cost-efficient default for most tasks. It delivers strong quality at high speed and consumes less quota than 2.5 Pro — making it a great everyday model. Switch to 2.5 Pro for complex reasoning or quality-critical output.
## Context window advantage: practical examples
To put the 1M token context window in perspective:
* **Codebase review**: Load an entire mid-sized application codebase and ask for a comprehensive architecture review, bug analysis, or refactoring recommendations
* **Document comparison**: Feed multiple lengthy contracts, research papers, or reports and ask the model to compare, contrast, or synthesize across all of them
* **Long conversations**: Maintain context across very long sessions without hitting the limits that truncate history in smaller-context models
Very long contexts (hundreds of thousands of tokens) can sometimes affect response quality — the model's attention can become diluted across very large inputs. For best results with extremely long documents, consider narrowing to the most relevant sections when possible.
## Related
Gemini 2.5 Pro has built-in reasoning — learn when to use reasoning-capable models.
Compare with Claude's 200k context and extended thinking.
Compare with GPT-5, GPT-4o, and the o-series reasoning models.
Which Gemini models count against your monthly quota.