Use Reasoning Models
Use models that work through a problem before answering: pick one from the catalog, let nrouter/auto detect the ask, or use the Responses endpoint.
Reasoning models allocate dynamic test-time compute to plan, verify, and deliberate before generating answers for complex mathematics, code synthesis, and multi-step analytical tasks.
Invocation
Specify reasoning models directly by model ID or let nrouter/auto classify heavy reasoning tasks automatically:
curl https://api.nrouter.ai/v1/chat/completions \
-H "Authorization: Bearer $NROUTER_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model": "deepseek-r1", "messages": [{"role": "user", "content": "How much does the ball cost?"}]}'Direct requests serve the named model without substitution; browse current models on Models and the models reference. Use Responses for multi-step agent reasoning and Messages for thinking blocks. Settles via x-nr-request-cost and x-nr-model. Related guides: Tool Calling, Structured Output, and Image Input.
Frequently asked questions
Is there a reasoning effort setting?
No; select a reasoning model from the catalog when tasks demand deliberate test-time compute.
What does a reasoning request cost?
Exact list price for the serving model, including generated reasoning tokens, reported in x-nr-request-cost.
If nrouter/auto picks a reasoning model, could a fallback cost more?
Yes; fallbacks under nrouter/auto may differ in price, with the serving model always reported in x-nr-model.