AI Model Commands - NeuraTrade

The ai command manages AI provider integrations and model availability.

List AI Models

List all available AI models:

./bin/neuratrade ai models

Example Output

Available AI Models
==================
- gpt-4-turbo (openai): tools, vision
- gpt-4 (openai): tools
- gpt-3.5-turbo (openai): tools
- claude-3-opus (anthropic): tools, vision
- claude-3-sonnet (anthropic): tools, vision
- claude-3-haiku (anthropic): tools
- gemini-pro (google): tools, vision
- mistral-large (mistral): tools

Example Output (Backend Unreachable)

Available AI Models
==================
Error: Could not reach API: connection refused

Make sure the NeuraTrade backend is running:
  neuratrade gateway start

Or check your configuration:
  neuratrade config status

Model Capabilities

Each model shows its supported capabilities:

tools - Function calling / tool use
vision - Image understanding

How It Works

Calls /api/v1/ai/models on backend API
Backend queries configured AI providers
Returns list of available models with capabilities
CLI formats and displays results

List AI Providers

List configured AI providers:

./bin/neuratrade ai providers

Example Output

Available AI Providers
=====================
- OpenAI (openai) [active]
- Anthropic (anthropic) [active]
- Google AI (google) [inactive]
- Mistral AI (mistral) [active]

Provider Status

active - Provider configured with valid API key
inactive - Provider not configured or API key missing

Example Output (Backend Unreachable)

Available AI Providers
=====================
Error: Could not reach API: connection refused

Make sure the NeuraTrade backend is running:
  neuratrade gateway start

Configuration

In `config.json`

Configure AI providers in the configuration file:

{
  "ai": {
    "provider": "openai",
    "api_key": "sk-...",
    "base_url": "https://api.openai.com/v1",
    "model": "gpt-4-turbo"
  }
}

Environment Variables

Override config with environment variables:

Variable	Description	Default
`AI_PROVIDER`	Default AI provider	From config
`AI_API_KEY`	Provider API key	From config
`AI_BASE_URL`	Provider API endpoint	Provider default
`AI_MODEL`	Default model	From config

Multiple Providers

Configure multiple providers for fallback or different use cases. The backend supports:

OpenAI - GPT models
Anthropic - Claude models
Google AI - Gemini models
Mistral AI - Mistral models
Custom Providers - OpenAI-compatible APIs

API Response Format

Models Endpoint

{
  "models": [
    {
      "id": "gpt-4-turbo",
      "display_name": "GPT-4 Turbo",
      "provider": "openai",
      "cost": "$0.01 / 1K tokens",
      "supports_tools": true,
      "supports_vision": true
    },
    {
      "id": "claude-3-opus",
      "display_name": "Claude 3 Opus",
      "provider": "anthropic",
      "cost": "$0.015 / 1K tokens",
      "supports_tools": true,
      "supports_vision": true
    }
  ]
}

Providers Endpoint

{
  "providers": [
    {
      "id": "openai",
      "name": "OpenAI",
      "is_active": true
    },
    {
      "id": "anthropic",
      "name": "Anthropic",
      "is_active": true
    },
    {
      "id": "google",
      "name": "Google AI",
      "is_active": false
    }
  ]
}

Model Selection

NeuraTrade automatically selects models based on task requirements:

Vision tasks - Uses models with supports_vision: true
Function calling - Uses models with supports_tools: true
Cost optimization - Selects cheaper models for simple tasks
Fallback - Tries alternative providers if primary fails

Checking Provider Status

Verify AI provider connectivity:

./bin/neuratrade health

Healthy output includes AI providers:

Health Check Results
===================
✓ Backend API: healthy

Service Health:
  ✓ database: healthy
  ✓ redis: healthy
  ✓ ai_provider_openai: healthy
  ✓ ai_provider_anthropic: healthy

Supported Providers

OpenAI

{
  "ai": {
    "provider": "openai",
    "api_key": "sk-...",
    "base_url": "https://api.openai.com/v1",
    "model": "gpt-4-turbo"
  }
}

Models:

gpt-4-turbo
gpt-4
gpt-3.5-turbo

Anthropic

{
  "ai": {
    "provider": "anthropic",
    "api_key": "sk-ant-...",
    "base_url": "https://api.anthropic.com",
    "model": "claude-3-opus-20240229"
  }
}

Models:

claude-3-opus-20240229
claude-3-sonnet-20240229
claude-3-haiku-20240307

Google AI

{
  "ai": {
    "provider": "google",
    "api_key": "...",
    "model": "gemini-pro"
  }
}

Models:

gemini-pro
gemini-pro-vision

Mistral AI

{
  "ai": {
    "provider": "mistral",
    "api_key": "...",
    "base_url": "https://api.mistral.ai/v1",
    "model": "mistral-large-latest"
  }
}

Models:

mistral-large-latest
mistral-medium-latest
mistral-small-latest

Custom OpenAI-Compatible Providers

{
  "ai": {
    "provider": "openai",
    "api_key": "...",
    "base_url": "https://your-custom-endpoint.com/v1",
    "model": "your-model-name"
  }
}

Works with:

LM Studio
LocalAI
Ollama (with OpenAI compatibility)
Azure OpenAI
Any OpenAI-compatible API

Security Best Practices

API Key Security:

Never commit API keys to version control
Use environment variables in production
Rotate keys regularly
Monitor usage for anomalies
Set spending limits in provider dashboards

Key Masking

The CLI and backend automatically mask API keys in logs and output:

AI Provider: OpenAI
API Key: sk-...***...xyz (masked)

File Permissions

Protect config files:

chmod 600 ~/.neuratrade/config.json

Troubleshooting

”No AI models available”

Ensure at least one provider is configured:

./bin/neuratrade config show

Add provider configuration:

./bin/neuratrade config init --ai-key YOUR_API_KEY

“Could not reach API”

Start the backend:

./bin/neuratrade gateway start

Verify backend is running:

./bin/neuratrade status

“Provider inactive”

Check API key configuration:

cat ~/.neuratrade/config.json | jq '.ai'

Verify API key is valid:

Log in to provider dashboard
Check API key status
Verify key has required permissions
Check spending limits and quotas

Rate Limits

If you hit rate limits:

Configure multiple providers for fallback
Increase delay between requests
Upgrade provider plan
Use cheaper models for non-critical tasks

Cost Monitoring

Monitor AI provider costs:

Check provider dashboard
Set up spending alerts
Review ai_model fields in logs
Use cheaper models for testing

NeuraTrade automatically selects cost-effective models based on task complexity. Override with the AI_MODEL environment variable if needed.

Commands

​List AI Models

​Example Output

​Example Output (Backend Unreachable)

​Model Capabilities

​How It Works

​List AI Providers

​Example Output

​Provider Status

​Example Output (Backend Unreachable)

​Configuration

​In config.json

​Environment Variables

​Multiple Providers

​API Response Format

​Models Endpoint

​Providers Endpoint

​Model Selection

​Checking Provider Status

​Supported Providers

​OpenAI

​Anthropic

​Google AI

​Mistral AI

​Custom OpenAI-Compatible Providers

​Security Best Practices

​Key Masking

​File Permissions

​Troubleshooting

​”No AI models available”

​“Could not reach API”

​“Provider inactive”

​Rate Limits

​Cost Monitoring

Build docs developers (and LLMs) love

List AI Models

Example Output

Example Output (Backend Unreachable)

Model Capabilities

How It Works

List AI Providers

Example Output

Provider Status

Example Output (Backend Unreachable)

Configuration

In `config.json`

Environment Variables

Multiple Providers

API Response Format

Models Endpoint

Providers Endpoint

Model Selection

Checking Provider Status

Supported Providers

OpenAI

Anthropic

Google AI

Mistral AI

Custom OpenAI-Compatible Providers

Security Best Practices

Key Masking

File Permissions

Troubleshooting

”No AI models available”

“Could not reach API”

“Provider inactive”

Rate Limits

Cost Monitoring