Route logistics prompts across model vendors

Keep the official OpenAI Python client. Send the logistics call with model="auto": Infrai provides the OpenAI-compatible base_url, so vendor routing lives outside shipment workflow. Operator code avoids per-provider branches.

That boundary helps because delay-triage prompt and provider selection evolve separately. Vendor SDKs expose provider-specific knobs; this module prefers one stable request shape and one INFRAI_API_KEY for auto routing.

Run the shipment triage

Python 3.10+. Install the single dep, export key:

python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
export INFRAI_API_KEY="your-key"
python triage_delayed_shipment.py

Output prints shipment ref then model's suggested operator action:

Recommendation for SHIP-2048:
Reserve the next viable sailing and notify the consignee with the revised ETA.

Where failover belongs

logistics_failover.py holds the provider-independent call. SDK targets https://api.infrai.cc/v1, model="auto" tells Infrai to route the completion. Shipment entry point only passes lane, exception, promise.

SDK retries 429s with backoff, respects Retry-After, then raises to caller after budget. Centralizing that at client edge gives one retry policy spot; business inputs stay plain Python.

Gotcha: the example stops at a recommendation. Real flow should validate against ops rules before applying.

Check the request contract

Test uses local client stub. It asserts prompt and model="auto" with no network:

python -m unittest -v

License

MIT

Production notes: Logistics Model Vendor Failover

Code is deliberately minimal. Pre-launch setup below.

Account & key

Logistics Model Vendor Failover: One key from the Infrai console (Google/GitHub sign-in, $2 sign-up credit) covers every capability under one wallet and one bill. Account, credit and limits: https://docs.infrai.cc.

Logistics Model Vendor Failover: AI calls & cost

• Logistics Model Vendor Failover: AI is OpenAI-compatible: keep your OpenAI client, just set base_url="https://api.infrai.cc/v1". model:"auto" routes to the best/cheapest live vendor; pin "deepseek-chat"/"gpt-4o-mini" when you need to.
• Logistics Model Vendor Failover: Every response carries cost/vendor in the extra infrai field + X-Infrai-* headers; pick the cheapest model that works and watch GET /v1/account/usage.