Reasoning models think more deeply before answering — good for math, code, and multi-step tasks. Use reasoning_effort to control depth (e.g. low / medium / high).
curl https://tokendog.io/v1/chat/completions \
-H "Authorization: Bearer $TOKENDOG_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model":"gpt-5.1",
"reasoning_effort":"high",
"messages":[{"role":"user","content":"A number plus half of itself equals 36. Find the number."}]
}'
Higher effort means deeper thinking but more latency and cost; use low for simple tasks.
