# ---- Required (at least one provider) ----
# vast.ai API key — Account → API Keys at https://cloud.vast.ai/
VAST_API_KEY=

# RunPod API key — Settings → API Keys at https://www.runpod.io/console/user/settings
# (only needed if you use --provider runpod)
RUNPOD_API_KEY=

# ---- Optional ----
# HuggingFace token — ONLY needed for gated/private models (Llama, Gemma, etc.)
# Also raises the rate limit on model-metadata lookups. Public models work without it.
HF_TOKEN=

# Bearer token that protects your vLLM endpoint on the public internet.
# Leave blank to have `aiod` generate a random one per launch.
VLLM_API_KEY=

# ---- Defaults (override per-run on the CLI) ----
# Auto-destroy the instance after this many hours so a forgotten box can't bill all weekend.
AIOD_TTL_HOURS=4
# Hard cap on price; offers above this are never rented.
AIOD_MAX_PRICE=6.0