Skip to Content
API GuidesResponses APIReasoning

Reasoning

Responses API Beta

The Responses API Beta supports advanced reasoning capabilities, allowing models to show their internal reasoning process with configurable effort levels.

Reasoning Configuration

Configure reasoning behavior using the reasoning parameter:

responses-reasoning-01.js
const response = await fetch("https://novapai.ai/api/v1/responses", {
  method: "POST",
  headers: {
    Authorization: "Bearer YOUR_NOVASTACK_AI_API_KEY",
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    model: "openai/o4-mini",
    input: "What is the meaning of life?",
    reasoning: {
      effort: "high",
    },
    max_output_tokens: 9000,
  }),
});
 
const result = await response.json();
console.log(result);

Reasoning Effort Levels

The effort parameter controls how much computational effort the model puts into reasoning:

Effort LevelDescription
minimalBasic reasoning with minimal computational effort
lowLight reasoning for simple problems
mediumBalanced reasoning for moderate complexity
highDeep reasoning for complex problems

Complex Reasoning Example

For complex mathematical or logical problems:

responses-reasoning-02.js
const response = await fetch("https://novapai.ai/api/v1/responses", {
  method: "POST",
  headers: {
    Authorization: "Bearer YOUR_NOVASTACK_AI_API_KEY",
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    model: "openai/o4-mini",
    input: [
      {
        type: "message",
        role: "user",
        content: [
          {
            type: "input_text",
            text: "Was 1995 30 years ago? Please show your reasoning.",
          },
        ],
      },
    ],
    reasoning: {
      effort: "high",
    },
    max_output_tokens: 9000,
  }),
});
 
const result = await response.json();
console.log(result);

Reasoning in Conversation Context

Include reasoning in multi-turn conversations:

responses-reasoning-03.js
const response = await fetch("https://novapai.ai/api/v1/responses", {
  method: "POST",
  headers: {
    Authorization: "Bearer YOUR_NOVASTACK_AI_API_KEY",
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    model: "openai/o4-mini",
    input: [
      {
        type: "message",
        role: "user",
        content: [
          {
            type: "input_text",
            text: "What is your favorite color?",
          },
        ],
      },
      {
        type: "message",
        role: "assistant",
        id: "msg_abc123",
        status: "completed",
        content: [
          {
            type: "output_text",
            text: "I don't have a favorite color.",
            annotations: [],
          },
        ],
      },
      {
        type: "message",
        role: "user",
        content: [
          {
            type: "input_text",
            text: "How many Earths can fit on Mars?",
          },
        ],
      },
    ],
    reasoning: {
      effort: "high",
    },
    max_output_tokens: 9000,
  }),
});
 
const result = await response.json();
console.log(result);

Streaming Reasoning

Enable streaming to see reasoning develop in real-time:

responses-reasoning-04.js
const response = await fetch("https://novapai.ai/api/v1/responses", {
  method: "POST",
  headers: {
    Authorization: "Bearer YOUR_NOVASTACK_AI_API_KEY",
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    model: "openai/o4-mini",
    input:
      "Solve this step by step: If a train travels 60 mph for 2.5 hours, how far does it go?",
    reasoning: {
      effort: "medium",
    },
    stream: true,
    max_output_tokens: 9000,
  }),
});
 
const reader = response.body?.getReader();
const decoder = new TextDecoder();
 
while (true) {
  const { done, value } = await reader.read();
  if (done) break;
 
  const chunk = decoder.decode(value);
  const lines = chunk.split("\n");
 
  for (const line of lines) {
    if (line.startsWith("data: ")) {
      const data = line.slice(6);
      if (data === "[DONE]") return;
 
      try {
        const parsed = JSON.parse(data);
        if (parsed.type === "response.reasoning.delta") {
          console.log("Reasoning:", parsed.delta);
        }
      } catch (e) {
        // Skip invalid JSON
      }
    }
  }
}

Response with Reasoning

When reasoning is enabled, the response includes reasoning information:

responses-reasoning-05.json
{
  "id": "resp_1234567890",
  "object": "response",
  "created_at": 1234567890,
  "model": "openai/o4-mini",
  "output": [
    {
      "type": "reasoning",
      "id": "rs_abc123",
      "encrypted_content": "gAAAAABotI9-FK1PbhZhaZk4yMrZw3XDI1AWFaKb9T0NQq7LndK6zaRB...",
      "summary": [
        "First, I need to determine the current year",
        "Then calculate the difference from 1995",
        "Finally, compare that to 30 years"
      ]
    },
    {
      "type": "message",
      "id": "msg_xyz789",
      "status": "completed",
      "role": "assistant",
      "content": [
        {
          "type": "output_text",
          "text": "Yes. In 2025, 1995 was 30 years ago. In fact, as of today (Aug 31, 2025), it's exactly 30 years since Aug 31, 1995.",
          "annotations": []
        }
      ]
    }
  ],
  "usage": {
    "input_tokens": 15,
    "output_tokens": 85,
    "output_tokens_details": {
      "reasoning_tokens": 45
    },
    "total_tokens": 100
  },
  "status": "completed"
}

Best Practices

.   **Choose appropriate effort levels**: Use `high` for complex problems, `low` for simple tasks
.   **Consider token usage**: Reasoning increases token consumption
.   **Use streaming**: For long reasoning chains, streaming provides better user experience
.   **Include context**: Provide sufficient context for the model to reason effectively

Next Steps

Basic UsageTool Calling