> ## Documentation Index
> Fetch the complete documentation index at: https://docs.enconvo.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Model Selection Guide

> Choose the right AI model for your specific use case

## Overview

EnConvo supports 25+ AI providers and hundreds of models. This guide helps you choose the right model for your specific needs, balancing quality, speed, cost, and privacy.

## Quick Recommendation

Not sure where to start? Here are our top picks for common scenarios:

| Scenario              | Recommended Model          | Why                                       |
| --------------------- | -------------------------- | ----------------------------------------- |
| **General daily use** | Claude Haiku 4.5           | Fast, affordable, handles most tasks well |
| **Complex reasoning** | Claude Sonnet 4.5 or GPT-5 | Deep analysis, multi-step problem solving |
| **Coding tasks**      | Claude Sonnet 4.5          | Best code understanding and generation    |
| **Creative writing**  | GPT-5 or Claude Opus 4.8   | Natural language, creative expression     |
| **Long documents**    | Gemini 2.5 Flash           | 1M token context window                   |
| **Math and science**  | DeepSeek V3                | Specialized reasoning capabilities        |
| **Privacy-first**     | Ollama (Llama 3)           | Runs entirely on your Mac                 |
| **Budget-friendly**   | Gemini 2.5 Flash Lite      | Excellent quality at lowest cost          |

## Model Capability Comparison

### Flagship Models

| Model                 | Provider                | Context         | Multimodal | Tool Use | Strengths                                           |
| --------------------- | ----------------------- | --------------- | :--------: | :------: | --------------------------------------------------- |
| **GPT-5**             | OpenAI                  | 128K            |     Yes    |    Yes   | Broad knowledge, creative tasks, vision             |
| **Claude Sonnet 4.5** | Anthropic               | 200K            |     Yes    |    Yes   | Coding, analysis, instruction following             |
| **Claude Opus 4.8**   | Anthropic               | 200K            |     Yes    |    Yes   | Complex reasoning, agentic tasks                    |
| **Gemini 2.5 Pro**    | Google                  | 1M              |     Yes    |    Yes   | Long context, research, multimodal                  |
| **Gemini 2.5 Flash**  | Google                  | 1M              |     Yes    |    Yes   | Fast, efficient, long context                       |
| **DeepSeek V3**       | DeepSeek                | 128K            |     Yes    |    Yes   | Math, coding, reasoning at low cost                 |
| **Grok**              | xAI                     | 128K            |     Yes    |    Yes   | Real-time knowledge, reasoning                      |
| **GLM-5.2**           | GLM / EnConvo Cloud     | Model-dependent |     Yes    |    Yes   | General chat, coding, Chinese and multilingual work |
| **MiniMax M3**        | MiniMax / EnConvo Cloud | Model-dependent |     Yes    |    Yes   | Fast general-purpose chat and writing               |

### Cost-Efficient Models

| Model                     | Provider  | Context | Best For                      | Relative Cost |
| ------------------------- | --------- | ------- | ----------------------------- | :-----------: |
| **GPT-5 Mini**            | OpenAI    | 128K    | General tasks, fast responses |      Low      |
| **Claude Haiku 4.5**      | Anthropic | 200K    | Quick tasks, high volume      |      Low      |
| **Gemini 2.5 Flash Lite** | Google    | 1M      | Budget-friendly, long context |     Lowest    |
| **DeepSeek V3**           | DeepSeek  | 128K    | Coding, math at great value   |      Low      |
| **Groq (Llama 3)**        | Groq      | 128K    | Ultra-fast inference          |      Low      |

### Local Models (Free, Private)

| Model                 | Platform  | RAM Needed | Best For                 |
| --------------------- | --------- | :--------: | ------------------------ |
| **Llama 3.3 70B**     | Ollama    |   40+ GB   | Most capable local model |
| **Llama 3.1 8B**      | Ollama    |    8 GB    | General tasks, fast      |
| **Qwen 2.5 72B**      | Ollama    |   40+ GB   | Multilingual, coding     |
| **Mistral Small**     | Ollama    |    8 GB    | European languages       |
| **CodeLlama 34B**     | Ollama    |   20+ GB   | Code-focused tasks       |
| **DeepSeek Coder V2** | Ollama    |   16+ GB   | Coding at moderate size  |
| Any GGUF model        | LM Studio |   Varies   | Custom model loading     |

## Choosing by Use Case

### Coding and Development

<Tabs>
  <Tab title="Best Overall">
    **Claude Sonnet 4.5** is the top choice for coding tasks:

    * Excellent code understanding across 20+ languages
    * Generates clean, well-structured code
    * Strong at debugging and refactoring
    * Understands complex codebases and architectures
    * 200K context for large code files
  </Tab>

  <Tab title="Best Value">
    **DeepSeek V3** offers exceptional coding ability at a fraction of the cost:

    * Competitive with top models on coding benchmarks
    * Strong mathematical reasoning
    * Very affordable API pricing
    * Good for high-volume coding tasks
  </Tab>

  <Tab title="Best Local">
    **Qwen 2.5 Coder** or **DeepSeek Coder V2** via Ollama:

    * No API costs
    * Complete privacy -- code never leaves your Mac
    * Good for sensitive/proprietary code
    * Requires Apple Silicon Mac with adequate RAM
  </Tab>
</Tabs>

### Writing and Content Creation

<Tabs>
  <Tab title="Creative Writing">
    **GPT-5** excels at creative tasks:

    * Natural, engaging prose
    * Strong narrative structure
    * Good at maintaining voice and tone
    * Excellent vocabulary and style variety

    **Claude Opus 4.8** is also excellent for nuanced, long-form writing.
  </Tab>

  <Tab title="Technical Writing">
    **Claude Sonnet 4.5** for technical documentation:

    * Precise, clear explanations
    * Consistent formatting
    * Good at adapting complexity to audience
    * Strong at structured documents
  </Tab>

  <Tab title="Quick Drafts">
    **Claude Haiku 4.5** or **GPT-5 Mini** for:

    * Email drafts
    * Quick summaries
    * Social media posts
    * Short-form content
    * Fast turnaround is more important than perfection
  </Tab>
</Tabs>

### Research and Analysis

<Tabs>
  <Tab title="Long Documents">
    **Gemini 2.5 Pro** or **Gemini 2.5 Flash** with their 1M token context:

    * Process entire books, research papers, or codebases
    * Cross-reference information across hundreds of pages
    * Enable Google Search grounding for real-time information
    * URL Context tool for analyzing web pages directly
  </Tab>

  <Tab title="Deep Analysis">
    **Claude Opus 4.8** for complex analytical tasks:

    * Multi-step reasoning chains
    * Nuanced interpretation of ambiguous data
    * Strong critical thinking
    * Good at identifying patterns and insights
  </Tab>

  <Tab title="Quick Lookups">
    **Perplexity** models for research with citations:

    * Built-in web search
    * Sources and citations included
    * Fast factual answers
    * Good for current events and recent information
  </Tab>
</Tabs>

### Math, Science, and Reasoning

| Task                  | Recommended Model          | Notes                                    |
| --------------------- | -------------------------- | ---------------------------------------- |
| **Advanced math**     | DeepSeek V3                | Strong mathematical reasoning            |
| **Physics/Chemistry** | Claude Sonnet 4.5 or GPT-5 | Good scientific knowledge                |
| **Data analysis**     | Gemini 2.5 Pro             | Handles large datasets with long context |
| **Logic puzzles**     | Claude Opus 4.8            | Excels at complex reasoning chains       |
| **Statistics**        | DeepSeek V3                | Cost-effective for quantitative tasks    |

### Multilingual Tasks

| Language Group      | Best Cloud Model   | Best Local Model |
| ------------------- | ------------------ | ---------------- |
| **English**         | Any top-tier model | Llama 3.1 8B     |
| **Chinese**         | DeepSeek V3, Qwen  | Qwen 2.5         |
| **Japanese/Korean** | GPT-5, Claude      | Qwen 2.5         |
| **European**        | Claude, GPT-5      | Mistral          |
| **Multilingual**    | Gemini 2.5         | Qwen 2.5 72B     |

## Cloud vs Local: Decision Guide

### Choose Cloud Models When:

* You need the highest accuracy and capability
* You work with many different tasks throughout the day
* You have reliable internet access
* Speed of response matters more than privacy
* You need the latest knowledge and capabilities

### Choose Local Models When:

* Privacy is critical (sensitive data, proprietary code)
* You work offline frequently
* You want zero ongoing API costs
* You have an Apple Silicon Mac with sufficient RAM
* Your tasks are well-suited to smaller models

### Hybrid Approach (Recommended)

Most users benefit from combining both:

| Task Type              | Use                              |
| ---------------------- | -------------------------------- |
| Sensitive code review  | Local (Ollama)                   |
| Creative writing       | Cloud (GPT-5 / Claude)           |
| Quick translations     | Cloud (fast model)               |
| Confidential documents | Local (Ollama)                   |
| Complex research       | Cloud (Gemini with long context) |
| Daily chat             | Cloud (Haiku / GPT-5 Mini)       |

## Multimodal Capabilities

Models that can understand images, audio, and other media:

| Model                 | Images | Audio | Video | Documents |
| --------------------- | :----: | :---: | :---: | :-------: |
| **GPT-5**             |   Yes  |  Yes  |   No  |    Yes    |
| **Claude Sonnet 4.5** |   Yes  |   No  |   No  |    Yes    |
| **Gemini 2.5 Pro**    |   Yes  |  Yes  |  Yes  |    Yes    |
| **Gemini 2.5 Flash**  |   Yes  |  Yes  |  Yes  |    Yes    |
| **Grok**              |   Yes  |   No  |   No  |    Yes    |

<Tip>
  If you frequently work with images (screenshots, diagrams, photos), choose a model with vision capabilities. Gemini models have the broadest multimodal support including video understanding.
</Tip>

## Cost Optimization Strategies

<AccordionGroup>
  <Accordion title="Use tiered models">
    Route simple tasks to cheap models and complex tasks to powerful ones:

    * **Quick questions:** Claude Haiku 4.5 or GPT-5 Mini
    * **Standard work:** Claude Sonnet 4.5 or GPT-5
    * **Complex analysis:** Claude Opus 4.8 or Gemini 2.5 Pro
  </Accordion>

  <Accordion title="Leverage Enconvo Cloud Plan">
    The Enconvo Cloud Plan provides access to all major providers through a single subscription with points-based pricing. This avoids managing multiple API keys and lets you switch models freely.
  </Accordion>

  <Accordion title="Use local models for high-volume tasks">
    If you run hundreds of queries daily, local models via Ollama eliminate per-query costs entirely. A one-time hardware investment pays for itself quickly.
  </Accordion>

  <Accordion title="Optimize context length">
    Sending unnecessary context increases costs. Use EnConvo's context awareness features to include only relevant information, and leverage knowledge base RAG instead of stuffing entire documents into the prompt.
  </Accordion>

  <Accordion title="Choose the right model size">
    Bigger is not always better. For simple tasks (formatting, extraction, classification), a small fast model delivers identical results at a fraction of the cost of a flagship model.
  </Accordion>
</AccordionGroup>

## Configuring Models in EnConvo

<Steps>
  <Step title="Open Settings">
    Navigate to any command that uses an AI model (e.g., AI Chat, Translator, Agent).
  </Step>

  <Step title="Select AI Model Provider">
    Find the **AI Model Provider** setting and click to change it. You will see all configured providers.
  </Step>

  <Step title="Choose Provider and Model">
    Select the provider (OpenAI, Anthropic, Google, etc.) and then choose the specific model from the **Model Name** dropdown.
  </Step>

  <Step title="Adjust Temperature">
    Set the temperature based on your needs:

    * **0 (none):** Deterministic, consistent responses -- best for coding, math, extraction
    * **0.5 (low):** Slightly varied -- good for general tasks
    * **1.0 (medium):** Balanced creativity -- default for most use cases
    * **1.5 (high):** More creative -- good for brainstorming, writing
    * **2.0 (maximum):** Most varied -- experimental, creative exploration
  </Step>
</Steps>

## Provider Setup Quick Reference

| Provider          | Setup                 | API Key Source                                         |
| ----------------- | --------------------- | ------------------------------------------------------ |
| **Enconvo Cloud** | Built-in, no setup    | Enconvo subscription                                   |
| **OpenAI**        | Add API key           | [platform.openai.com](https://platform.openai.com)     |
| **Anthropic**     | Add API key           | [console.anthropic.com](https://console.anthropic.com) |
| **Google**        | Add API key           | [ai.google.dev](https://ai.google.dev)                 |
| **DeepSeek**      | Add API key           | [platform.deepseek.com](https://platform.deepseek.com) |
| **Groq**          | Add API key           | [console.groq.com](https://console.groq.com)           |
| **Ollama**        | Install Ollama app    | Free, no key needed                                    |
| **LM Studio**     | Install LM Studio app | Free, no key needed                                    |
| **OpenRouter**    | Add API key           | [openrouter.ai](https://openrouter.ai)                 |
| **Mistral**       | Add API key           | [console.mistral.ai](https://console.mistral.ai)       |
| **Perplexity**    | Add API key           | [perplexity.ai](https://perplexity.ai)                 |
| **xAI**           | Add API key           | [console.x.ai](https://console.x.ai)                   |

<Note>
  For detailed setup instructions for each provider, see the [Providers](/providers/overview) section of the documentation.
</Note>

## Switching Models On-the-Fly

In the AI Chat interface, you can switch models at any time:

1. Click the model name in the chat header
2. Select a different model from the dropdown
3. Continue your conversation with the new model

This is useful for:

* Starting with a fast model for quick back-and-forth, then switching to a powerful model for the final task
* Trying the same prompt with different models to compare results
* Using a vision model when you need to share an image

## Related Features

<CardGroup cols={2}>
  <Card title="Provider Setup" icon="gear" href="/providers/overview">
    Configure AI providers and API keys
  </Card>

  <Card title="Local LLM" icon="shield-halved" href="/ai/local-llm">
    Set up Ollama and local models
  </Card>

  <Card title="AI Chat" icon="comments" href="/ai/chat">
    Use models in conversation
  </Card>

  <Card title="Enconvo Cloud Plan" icon="cloud" href="/providers/enconvo">
    Access all models with one subscription
  </Card>
</CardGroup>