env.example

# Configure one or both upstream providers.

# MiniMax API key from https://platform.minimax.io
MINIMAX_API_KEY=your-minimax-api-key-here

# OpenAI API key from https://platform.openai.com/api-keys
# Needed only if you want the proxy to route some models directly to OpenAI.
# OPENAI_API_KEY=sk-...

# Optional: OpenRouter API key for web search support
# Only used for MiniMax requests that invoke Codex web_search.
OPENROUTER_API_KEY=your-openrouter-api-key-here

# Optional: GitHub token for authenticated API calls (prevents 404 on private repos & rate limits)
# GITHUB_TOKEN=ghp_your-github-token-here

# Optional: Override defaults
# PROXY_PORT=4000
# MINIMAX_BASE_URL=https://api.minimax.io/v1
# OPENAI_BASE_URL=https://api.openai.com/v1
# OPENROUTER_SEARCH_MODEL=nvidia/nemotron-3-super-120b-a12b:free
# DEFAULT_PROVIDER=minimax

# Optional: models the proxy should advertise from /v1/models
# MINIMAX_MODELS=MiniMax-M2.7
# OPENAI_MODELS=gpt-5.4,gpt-5.4-mini,gpt-5.4-nano,gpt-4o

# Optional: heuristic prefixes that should route to OpenAI even if not explicitly listed above
# OPENAI_MODEL_PREFIXES=gpt-,o1,o3,o4,codex-,chatgpt-