Ask one question

diffuse query is the one-shot form of chat. It discovers a model, routes through it, prints the answer, and exits. It is the right tool for scripts and pipelines.

diffuse query --model Qwen/Qwen2.5-0.5B-Instruct \
  --prompt "Explain black holes in two sentences."

Options

Flag	Default	Meaning
--model <id>	required	the model to query
--prompt <text>	required	the prompt to send
--bootstrap <urls>	built-in sentinels	sentinels to discover the network
--max-tokens <n>	80	maximum tokens to generate

Examples

A longer answer:

diffuse query --model Qwen/Qwen2.5-0.5B-Instruct \
  --prompt "Write a haiku about distributed systems." \
  --max-tokens 60

Against your own network:

diffuse query --model Qwen/Qwen2.5-0.5B-Instruct \
  --prompt "Ping" \
  --bootstrap http://your-sentinel:9440

Errors you may see

• Model not found on the network. No node serves that model. Run diffuse models to see what is live.
• Model is incomplete. The model is present but some layer range is held by nobody. The error names the missing slices. See replication.

Next

• Chat with the network
• OpenAI-compatible server