No completion on server, works on CLI

[etlweather]

I must be missing something dumb but I can't find it. Any pointer would be welcome.

llama-cli.exe -m c:\tmp\coedit-large-q4_k_m.gguf -p "Fix the grammar: [text removed]"

Works fine. The output on the console is the text with the grammar fixed.

docker run --rm -it -v c:\tmp:/models --gpus all -p 8000:8000 ghcr.io/ggml-org/llama.cpp:server-cuda -m /models/coedit-large-q4_k_m.gguf --port 8000 --host 0.0.0.0

POST http://localhost:8000/completion
Body:
{
  "prompt": "Fix the grammar: [text removed]"
}

Model loads but returns empty completion (content).