Skip to content

Configuration & API Keys

Goal

Make examples runnable in under 5 minutes with explicit model configuration.

Supported input methods

QitOS examples (via examples/common.py) support:

  1. CLI args
  2. Environment variables

Priority for API key is:

  1. --api-key
  2. OPENAI_API_KEY
  3. QITOS_API_KEY
  • --model-base-url default: https://api.siliconflow.cn/v1/
  • --model-name default: Qwen/Qwen3-8B
  • --temperature default: 0.2
  • --max-tokens default: 2048
export OPENAI_BASE_URL="https://api.siliconflow.cn/v1/"
export OPENAI_API_KEY="<your_api_key>"

Keep your API key out of git

Do not hardcode keys in examples or commit them to Git.

Recommended:

  • export them in your shell profile, or
  • use a local .env file and load it in your shell (do not commit .env)

Then run with explicit model endpoint:

python examples/patterns/react.py \
  --model-base-url "$OPENAI_BASE_URL" \
  --model-name "Qwen/Qwen3-8B" \
  --workspace ./playground

Using CLI-only config (no env vars)

python examples/patterns/react.py \
  --model-base-url "https://api.siliconflow.cn/v1/" \
  --api-key "<your_api_key>" \
  --model-name "Qwen/Qwen3-8B" \
  --workspace ./playground

Verify your config quickly

Run:

python examples/patterns/react.py --max-steps 1 --workspace ./playground

If config is correct, you should see:

  1. model call activity in render output
  2. no Missing API key exception
  3. trace artifacts in runs/ (unless --disable-trace)

Common errors

  1. Missing API key...

  2. set --api-key or OPENAI_API_KEY/QITOS_API_KEY

  3. request 401/403

  4. key invalid or endpoint does not accept this key

  5. request 404/model not found

  6. wrong --model-name for that provider

  7. timeout

  8. endpoint unavailable, network/proxy issue, or too strict timeout in provider

Source Index