AI Models

Choose the right intelligence for your developer workflows in M0X CoderX and M0X IDE.

M0X abstracts away the complexities of different AI providers, offering a unified, high-performance model interface. You can switch seamlessly between premium cloud LLMs directly within the editor to optimize for reasoning, speed, or autocomplete latency.

☁️ Supported Cloud Models

Cloud-hosted models provide the highest reasoning capability, code accuracy, and speed for complex tasks. M0X natively supports the following APIs:

1. Anthropic Claude Family

Claude 3.5 Sonnet (Recommended)
- Context Window: 200k tokens.
- Best For: Flagship reasoning, multi-file code editing, agentic tool use, and debugging.
- M0X Role: The default recommended model for CoderX AGENT mode and M0X IDE Chat (Ctrl+L). It excels at writing syntactically correct code and editing complex component patterns.
Claude 3.5 Haiku
- Context Window: 200k tokens.
- Best For: Speed, lightweight refactors, and simple explanations.
- M0X Role: Ideal for fast inline edits (Ctrl+K) and real-time completions.

2. OpenAI GPT Family

GPT-4o
- Context Window: 128k tokens.
- Best For: Multi-turn conversational planning, explanations, and creative writing.
- M0X Role: High-speed premium chat assistant in both CoderX and IDE.
GPT-4o-mini
- Context Window: 128k tokens.
- Best For: Fast, low-latency logic checks and simple code rewrites.
- M0X Role: Lightweight editor helper.

3. Google Gemini Family

Gemini 1.5 Pro
- Context Window: 2 Million tokens.
- Best For: Full repository analysis, multi-hour logs ingest, and massive codebases.
- M0X Role: If you are trying to write a complex feature that spans an entire workspace, Gemini 1.5 Pro allows you to feed your entire project context into CoderX without overflowing the context limits.
Gemini 1.5 Flash
- Context Window: 1 Million tokens.
- Best For: Speedy general coding and summarization.

4. DeepSeek Family

DeepSeek-V3
- Context Window: 128k tokens.
- Best For: High-performance reasoning at ultra-low cost.
- M0X Role: State-of-the-art general coding and chat.
DeepSeek-R1
- Context Window: 128k tokens.
- Best For: Complex algorithmic logic and step-by-step mathematical reasoning.

⚙️ Model Allocation & Configuration

M0X allows you to assign specialized models to specific editor tasks. This balances speed, cost, and reasoning power.

Navigate to Settings > Models to configure:

Task Area	Recommended Model	Reason
Workspace Chat / Planning	`Claude 3.5 Sonnet` / `GPT-4o`	Requires high reasoning for multi-step tasks.
Agentic Loop (Executions)	`Claude 3.5 Sonnet`	Tool-use accuracy is highest in Sonnet.
Inline Edit (`Ctrl+K`)	`Claude 3.5 Sonnet` / `GPT-4o`	Requires fast response with high syntax accuracy.
Autocomplete (Ghost Text)	`DeepSeek-V3` / `GPT-4o-mini`	Balanced low latency with high token match precision.

Adding Custom API Endpoints

If your team uses a secure proxy, Azure OpenAI, or a self-hosted server, you can input a custom endpoint URL:

Go to Settings > API Keys.
Select your provider, check Use Custom Endpoint, and paste your base URL (e.g. https://gateway.internal.corp/v1).
Save settings to verify the connection.