Roo Code

Overview

Featherless AI provides seamless access to a vast catalog of high-performance open-source models, enabling you to power Roo Code with cutting-edge AI capabilities. Choose from specialized coding models, general-purpose LLMs, or reasoning-focused models - all through a single, OpenAI-compatible API.

Key Benefits:

• 🚀 Extensive Model Catalog - Access dozens of open-source models without managing infrastructure

• 💰 Flexible Pricing - Various subscription tiers to match your usage needs

• ⚡ No Setup Required - Models are ready to use immediately, no deployment needed

• 🎯 Specialized Options - Choose task-specific models for optimal performance

Prerequisites

Before you begin, ensure you have:

• Visual Studio Code installed

• Roo Code extension installed and activated

• A Featherless AI account (view pricing)

Getting Started

Step 1: Obtain Your API Key

1. Visit Featherless AI: Navigate to https://featherless.ai

2. Create or Sign Into Your Account: Click "Sign Up" if you're new, or "Sign In" for existing users

3. Access API Keys Section: Once logged in, go to your account dashboard and find the API Keys section

4. Generate New Key:

   • Click "Create New API Key"

   • Give it a descriptive name (e.g., "Roo Code Integration")

   • Click "Generate"

5. Save Your Key: ⚠️ Important: Copy the API key immediately and store it securely. For security reasons, it will only be displayed once.

Step 2: Choose Your Model

Visit the Featherless Model Catalog to explore available options. Each model page displays:

• Model ID (needed for configuration)

• Context window size

• Capabilities and specializations

• Token usage rates

For subscription details and pricing, visit the Featherless Pricing Page.

Configuration

Setting Up Roo Code

1. Open Roo Code Settings

   • Click the Roo Code icon in the VS Code sidebar

   • Click the settings gear icon (⚙️) in the Roo Code panel

2. Configure API Connection

   Setting	Value
   API Provider	Featherless AI
   OpenAI API Key	Your Featherless API key
   Model	Model ID from catalog (e.g., deepseek-ai/DeepSeek-V3-0324)

3. Save and Test

   • Click "Save" to apply settings

   • Test with a simple prompt to verify connection

Supported Models

Featured Models

DeepSeek Models

• deepseek-ai/DeepSeek-R1-0528 🧠

  • Advanced reasoning with <think> tag support

  • Ideal for complex problem-solving

  • Context: ~32K tokens

• deepseek-ai/DeepSeek-V3-0324

  • General-purpose powerhouse

  • Excellent for varied coding tasks

  • Context: ~32K tokens

Instruction-Following Models

• moonshotai/Kimi-K2-Instruct

  • Balanced general-purpose model

  • Good at following complex instructions

  • Context: ~32K tokens

• openai/gpt-oss-120b

  • Large parameter count for nuanced understanding

  • Versatile across different tasks

  • Context: ~32K tokens

Model Selection Guide

Advanced Configuration

Optimizing Performance

1. Context Window Management

   • Set context size appropriately for your project size

   • Larger contexts = better understanding but slower responses

   • Start with default and adjust based on needs

2. Model-Specific Settings

   • For Reasoning Models: Enable R1 parameters

   • For Code Tasks: Consider specialized coding models

   • For Quick Edits: Use smaller, faster models

Troubleshooting

Common Issues and Solutions

Debug Checklist

1. API key is valid and active

2. Base URL is exactly https://api.featherless.ai/v1

3. Model ID matches catalog listing exactly

4. Context window doesn't exceed model limits

5. Network connection is stable

6. Roo Code extension is up to date

Best Practices

1. Model Selection Strategy

• Start with general-purpose models for exploration

• Switch to specialized models for specific tasks

• Use reasoning models for complex architectural decisions

2. Prompt Engineering

• Be specific and provide context

• Include examples when possible

• Break complex tasks into steps

4. Performance Optimization

• Cache frequently used completions

• Batch similar requests when possible

• Use appropriate model sizes for tasks

FAQ

Q: What subscription plans are available? A: Featherless offers various subscription tiers to match different usage needs. Visit the pricing page for current plans and details.

Q: Can I switch models mid-session? A: Yes! Simply update the model ID in settings and save. The change takes effect immediately.

Q: Do models support image inputs? A: Currently, image support is not available through the Featherless integration.

Q: How do I know which model to choose? A: Start with recommended models above, then experiment based on your specific needs. The model catalog provides detailed capabilities for each option.

Additional Resources