Skip to main content

Reasoning

To enable reasoning, specify reasoning_effort in your API request and use one of the supported values.

Notes​

  • OpenAI does not expose actual reasoning tokens; you will not see them in the response.
  • Deepseek reasoning models enable reasoning automatically; you do not need to set anything to turn it on.
  • Reasoning content in responses appears under reasoning_content.

Using Anthropic via OneRoute​

  • If the client specifies a reasoning_effort string ("low"/"medium"/"high"), OneRoute converts it to a numeric budget according to the table below.

Conversion table:

  • "low" -> 1024
  • "medium" -> 40968192
  • "high" -> 16384

Using Vertex AI via OneRoute​

  • If the client specifies a reasoning_effort string ("low"/"medium"/"high"/"none"), OneRoute converts it to a budget according to the table below.

Conversion table:

  • "none" -> 0
  • "low" -> 1024
  • "medium" -> 4096
  • "high" -> 16384

Reasoning Code Example​

import openai

client = openai.OpenAI(
    api_key="sk-37fbb5a2-989a-4364-ac81-e58281d31e13",
    base_url="https://gw.1route.ai/v1",
)

response = client.chat.completions.create(
    model="MiniMax-M2.1",
    messages=[
        {"role": "user", "content": "Hi, how are you?"}
    ],
    reasoning_effort="low"
)

print(response.choices[0].message.content)