Skip to main content
Glama
.env.example•1.6 kB
# FLUX MCP Server Configuration # Copy this file to .env and customize the values # Auto-unload timeout in seconds # The model will be automatically unloaded after this period of inactivity # Set to 0 to disable auto-unload # Default: 300 (5 minutes) FLUX_UNLOAD_TIMEOUT=300 # Output directory for generated images # Images will be saved as {timestamp}_{seed}.png # Default: ~/flux_output FLUX_OUTPUT_DIR=/path/to/flux_output # Optional: HuggingFace model cache directory # If not set, uses HuggingFace's default cache location # Useful for custom cache locations or shared model storage # FLUX_MODEL_CACHE=/path/to/model/cache # Offline mode: Disable HuggingFace API calls when model is cached # Set to 1 to work completely offline (recommended for privacy) # Default: 0 (online, checks for updates) HF_HUB_OFFLINE=1 # Model selection # Choose which FLUX model to use by default (both optimized for quality) # - black-forest-labs/FLUX.1-dev: Faster quality (4-8 min, 40 steps default) # - black-forest-labs/FLUX.2-dev: Maximum quality (30-40 min, 50 steps default) # Default: black-forest-labs/FLUX.2-dev # FLUX_MODEL_ID=black-forest-labs/FLUX.2-dev # Default inference steps for image generation # Model-specific smart defaults apply automatically: # - FLUX.1-dev: 40 steps (optimized for quality and speed) # - FLUX.2-dev: 50 steps (optimized for maximum quality) # You can override these with this env var # Default: 50 (FLUX.2-dev) # FLUX_DEFAULT_STEPS=50 # Default guidance scale for prompt adherence # Both models use 7.5 for optimal quality # Default: 7.5 # FLUX_DEFAULT_GUIDANCE=7.5

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/tehw0lf/flux-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server