Configuration

Settings File

Location: ~/.config/fastlane/settings.toml

[general]
host = "127.0.0.1"    # Bind address
port = 8787           # Port number
mode = "balanced"     # cheap, balanced, or accurate

[byok]
openai_key = "sk-..."
anthropic_key = "sk-ant-..."
groq_key = "gsk-..."
openrouter_key = "sk-or-..."
backend_url = "https://api.fastlane.dev"
device_id = "auto-generated-uuid"

Environment Variables

# Override model directory
export FASTLANE_MODEL_DIR=/path/to/models

# Rust logging
export RUST_LOG=fastlane=debug,tower_http=debug

CLI Options

fastlane [OPTIONS]

Options:
  --host <HOST>      Bind address [default: 127.0.0.1]
  --port <PORT>      Port number [default: 8787]
  --model-dir <DIR>  Path to ONNX model directory

Model Directory

The ONNX model files should be in:

training/models/baseline/
├── model.onnx
├── model.onnx.data
├── tokenizer.json
├── tokenizer_config.json
└── meta.json

By default, Fastlane looks in:

1. $FASTLANE_MODEL_DIR environment variable
2. ~/.local/share/fastlane/models/baseline/
3. training/models/baseline/ (relative to binary)