Providers

LeanCTL supports 7+ AI providers out of the box. Use local models for free, or connect your own cloud API keys with Pro.

Overview

Local Providers (Free)

Ollama

The easiest way to get started for free. Install Ollama, pull a model, and go.

# Install Ollama
# macOS: brew install ollama
# Linux: curl -fsSL https://ollama.ai/install.sh | sh

# Pull a model
ollama pull qwen3.5:27b       # recommended for coding
ollama pull gemma4:e2b         # Google's Gemma 4

# Configure via wizard
leanctl init

Or add directly to ~/.leanctl/config.toml:

[[providers]]
name = "local"
type = "ollama"
model = "qwen3.5:27b"
base_url = "http://localhost:11434"   # default

Remote Ollama

Connect to an Ollama instance running on another machine:

[[providers]]
name = "murat"
type = "ollama"
model = "qwen3.5:27b"
base_url = "http://192.168.1.100:11434"

LM Studio

LM Studio provides a GUI for downloading and running models. Start the local server, then configure:

[[providers]]
name = "lmstudio"
type = "lmstudio"
model = "qwen3.5-27b"
base_url = "http://localhost:1234"   # default

Cloud Providers (Pro)

Cloud providers require LeanCTL Pro ($9/month). You bring your own API key — LeanCTL does not proxy your requests.

Anthropic (Claude)

[[providers]]
name = "claude"
type = "anthropic"
model = "claude-sonnet-4-20250514"
api_key_env = "ANTHROPIC_API_KEY"

Supported models: claude-sonnet-4-20250514, claude-opus-4-20250514, and all Claude 3.x models.

Get your API key: console.anthropic.com

OpenAI (GPT)

[[providers]]
name = "gpt"
type = "openai"
model = "gpt-4.1"
api_key_env = "OPENAI_API_KEY"

Supported models: gpt-4.1, gpt-4.1-mini, o4-mini, and all OpenAI chat models.

Get your API key: platform.openai.com

DeepSeek

[[providers]]
name = "deepseek"
type = "deepseek"
model = "deepseek-chat"
api_key_env = "DEEPSEEK_API_KEY"
base_url = "https://api.deepseek.com/v1"

Get your API key: platform.deepseek.com

Groq

[[providers]]
name = "groq"
type = "groq"
model = "llama-3.3-70b-versatile"
api_key_env = "GROQ_API_KEY"
base_url = "https://api.groq.com/openai/v1"

Get your API key: console.groq.com

OpenRouter

[[providers]]
name = "openrouter"
type = "openrouter"
model = "anthropic/claude-sonnet-4"
api_key_env = "OPENROUTER_API_KEY"
base_url = "https://openrouter.ai/api/v1"

Access 100+ models through a single API. Get your key: openrouter.ai

Custom OpenAI-Compatible Endpoints

Any endpoint that implements the OpenAI chat completions API works with LeanCTL:

[[providers]]
name = "my-server"
type = "openai"
model = "my-model"
base_url = "https://my-server.com/v1"
api_key = "my-key"
local = true   # set true to bypass tier check for self-hosted

This works with: vLLM, text-generation-inference, LocalAI, llama.cpp server, and any other OpenAI-compatible API.

Switching Providers

Switch providers and models at any time:

In the TUI

/model                         # open model picker (Ctrl+L)
/model claude-sonnet-4-20250514    # switch model
/model claude/claude-sonnet-4  # switch provider + model
/connect list                  # show all configured providers
/connect test                  # test current connection

From the CLI

leanctl --model ollama/qwen3.5:27b
leanctl --model claude-sonnet-4-20250514
leanctl models

API Key Security

Best practices for managing API keys:

1. Environment variables (recommended): Set keys in your shell profile and reference them with api_key_env.
2. Config file: Store directly in config.toml with api_key. The file is in your home directory (~/.leanctl/), but is not encrypted.
3. .env file: LeanCTL loads .env from your project root. Add .env to .gitignore.

Free vs Pro

LeanCTL determines if a provider is "local" or "cloud" automatically:

• Local (Free): ollama type, lmstudio type, or any provider with local = true, or a base_url pointing to localhost, 127.0.0.1, or private IP ranges.
• Cloud (Pro): anthropic, openai, deepseek, groq, openrouter types without local override.

To upgrade to Pro:

leanctl register     # free beta registration
leanctl subscribe    # or subscribe at leanctl.com/#pricing
leanctl login        # enter your API key

Streaming & Usage Tracking

All providers support streaming responses. LeanCTL tracks token usage per turn:

View costs with /cost in the TUI or leanctl stats from the command line.

Context Windows

LeanCTL automatically detects the context window for known models:

Override with the context_window field in your provider definition if needed.