Supported LLM Providers

LEAI features a native, lightweight REST HTTP client, eliminating heavy external dependencies while connecting directly to industry-standard AI platforms.

🌐 Supported Providers & Environment Variables

Provider	`provider` in `leai.yml`	Environment Variable	Recommended Models
Ollama (Local / Free)	`ollama`	None (requires local Ollama server)	`qwen2.5-coder:latest`, `llama3.1`, `mistral`
Local (LM Studio / vLLM)	`local` or `custom`	None (requires local running server)	`qwen2.5`, `meta-llama-3.1-8b-instruct`
Google Gemini	`gemini`	`GEMINI_API_KEY`	`gemini-2.5-flash`, `gemini-1.5-pro`
OpenAI	`openai`	`OPENAI_API_KEY`	`gpt-4o`, `gpt-4o-mini`, `o3-mini`
Anthropic Claude	`claude` / `anthropic`	`ANTHROPIC_API_KEY`	`claude-3-5-sonnet-20241022`, `claude-3-5-haiku-20241022`
DeepSeek	`deepseek`	`DEEPSEEK_API_KEY`	`deepseek-chat`, `deepseek-reasoner`
Qwen (Alibaba)	`qwen`	`DASHSCOPE_API_KEY`	`qwen-plus`, `qwen-max`, `qwen-turbo`
Moonshot Kimi	`kimi`	`MOONSHOT_API_KEY`	`moonshot-v1-8k`, `moonshot-v1-32k`
Grok / xAI	`grok` or `xai`	`GROK_API_KEY` / `XAI_API_KEY`	`grok-2-latest`

⚙️ Configuration Examples in `leai.yml`

LEAI supports global defaults alongside granular per-provider overrides:

ai:
  default_provider: "ollama"      # Default active provider
  temperature: 0.2                # Global temperature (0.0 to 1.0)
  timeout: 300.0                  # Global timeout in seconds
  max_history_turns: 15           # History context memory window (turns)
  max_agent_iterations: 10        # Max tool calling steps per agent turn
  max_subagent_iterations: 5      # Max iterations for specialist subagents

  providers:
    # 1. Local Ollama
    ollama:
      base_url: "http://localhost:11434/v1"
      model: "qwen2.5-coder:latest"
      temperature: 0.1
      timeout: 300.0

    # 2. Local OpenAI-compatible server (LM Studio, vLLM, LocalAI)
    local:
      base_url: "http://localhost:1234/v1"
      model: "qwen2.5"
      temperature: 0.1

    # 3. Google Gemini
    gemini:
      api_key: "${GEMINI_API_KEY}"
      model: "gemini-2.5-flash"

    # 4. OpenAI
    openai:
      api_key: "${OPENAI_API_KEY}"
      model: "gpt-4o-mini"
      temperature: 0.2
      timeout: 120.0

🧪 Validating Models with `leai models`

Run the diagnostic command to test API connectivity and view available models:

```bash

List models for the default configured provider

leai models

List models for a specific provider

leai models -p gemini leai models -p ollama