The model hands one question to Parallel, which researches it on the web and returns a cited answer.
{"type": "parallel:research"} to tools on a chat request, or send the header x-tm-tools: parallel:research.curl https://api.routerplus.com/v1/chat/completions \
-H "Authorization: Bearer $TM_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "claude-haiku-4-5",
"messages": [{"role": "user", "content": "Which open-source databases added vector search in 2026?"}],
"tools": [{"type": "parallel:research", "parameters": {"max_uses": 1, "max_effort": "medium"}}]
}'
The price is the provider's list price with no markup, charged per call to your RouterPlus balance; the model's own turns are billed as ordinary chat requests.