DeepSeek V3.2 Speciale

This documentation is valid for the following list of our models:

• deepseek/deepseek-v3.2-speciale

Try in Playground

Model Overview

A high-compute variant of DeepSeek-V3.2 that outperforms GPT-5 and matches Gemini-3.0-Pro in reasoning benchmarks, achieving gold-medal-level results at the 2025 International Mathematical Olympiad (IMO) and the International Olympiad in Informatics (IOI).

Create AI/ML API Key

How to make the first API call

If you need a more detailed walkthrough for setting up your development environment and making a request step by step — feel free to use our Quickstart guide.

API Schema

post

Body

application/json

modelstring · enumRequiredPossible values:

max_completion_tokensinteger · min: 1Optional

An upper bound for the number of tokens that can be generated for a completion, including visible output tokens and reasoning tokens.

max_tokensnumber · min: 1Optional

The maximum number of tokens that can be generated in the chat completion. This value can be used to control costs for text generated via API.

streambooleanOptional

If set to True, the model response data will be streamed to the client as it is generated using server-sent events.

Default: false

temperaturenumber · max: 2Optional

What sampling temperature to use. Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic. We generally recommend altering this or top_p but not both.

top_pnumber · min: 0.01 · max: 1Optional

An alternative to sampling with temperature, called nucleus sampling, where the model considers the results of the tokens with top_p probability mass. So 0.1 means only the tokens comprising the top 10% probability mass are considered. We generally recommend altering this or temperature but not both.

stopany ofOptional

Up to 4 sequences where the API will stop generating further tokens. The returned text will not contain the stop sequence.

stringOptional

or

string[]Optional

or

any · nullableOptional

frequency_penaltynumber · min: -2 · max: 2 · nullableOptional

Number between -2.0 and 2.0. Positive values penalize new tokens based on their existing frequency in the text so far, decreasing the model's likelihood to repeat the same line verbatim.

presence_penaltynumber · min: -2 · max: 2 · nullableOptional

Positive values penalize new tokens based on whether they appear in the text so far, increasing the model's likelihood to talk about new topics.

seedinteger · min: 1Optional

This feature is in Beta. If specified, our system will make a best effort to sample deterministically, such that repeated requests with the same seed and parameters should return the same result.

echobooleanOptional

If True, the response will contain the prompt. Can be used with logprobs to return prompt logprobs.

min_pnumber · min: 0.001 · max: 0.999Optional

A number between 0.001 and 0.999 that can be used as an alternative to top_p and top_k.

top_knumberOptional

Only sample from the top K options for each subsequent token. Used to remove "long tail" low probability responses. Recommended for advanced use cases only. You usually only need to use temperature.

repetition_penaltynumber · nullableOptional

A number that controls the diversity of generated text by reducing the likelihood of repeated sequences. Higher values decrease repetition.

top_anumber · max: 1Optional

Alternate top sampling parameter.

response_formatone ofOptional

An object specifying the format that the model must output.

or

or

Responses

200Success

application/jsontext/event-stream

post

/v1/chat/completions

JavaScriptPythoncURLHTTP

async function main() {
  const response = await fetch('https://api.aimlapi.com/v1/chat/completions', {
    method: 'POST',
    headers: {
      'Authorization': 'Bearer <YOUR_AIMLAPI_KEY>',
      'Content-Type': 'application/json',
    },
    body: JSON.stringify({
      model: 'deepseek/deepseek-chat-v3.1',
      messages:[
          {
              role:'user',
              content: 'Hello'
          }
      ],
    }),
  });

  const data = await response.json();
  console.log(JSON.stringify(data, null, 2));
}

main();

200Success

{
  "id": "chatcmpl-CQ9FPg3osank0dx0k46Z53LTqtXMl",
  "object": "chat.completion",
  "created": 1762343744,
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "Hello! I'm just a program, so I don't have feelings, but I'm here and ready to help you. How can I assist you today?",
        "refusal": null,
        "annotations": null,
        "audio": null,
        "tool_calls": null
      },
      "finish_reason": "stop",
      "logprobs": null
    }
  ],
  "model": "deepseek/deepseek-chat-v3.1",
  "usage": {
    "prompt_tokens": 137,
    "completion_tokens": 914,
    "total_tokens": 1051,
    "completion_tokens_details": null,
    "prompt_tokens_details": null
  }
}

Code Example

Python

import requests
import json  # for getting a structured output with indentation 

response = requests.post(
    "https://api.aimlapi.com/v1/chat/completions",
    headers={
        # Insert your AIML API Key instead of <YOUR_AIMLAPI_KEY>:
        "Authorization":"Bearer <YOUR_AIMLAPI_KEY>",
        "Content-Type":"application/json"
    },
    json={
        "model":"deepseek/deepseek-v3.2-speciale",
        "messages":[
            {
                "role":"user",
                "content":"Hi! What do you think about mankind?" # insert your prompt
            }
        ]
    }
)

data = response.json()
print(json.dumps(data, indent=2, ensure_ascii=False))

JavaScript

Response

{
  "id": "gen-1770021770-coQRs5BE5oFW8jhEBDjN",
  "provider": "Parasail",
  "model": "deepseek/deepseek-v3.2-speciale",
  "object": "chat.completion",
  "created": 1770021770,
  "choices": [
    {
      "logprobs": null,
      "finish_reason": "stop",
      "native_finish_reason": "stop",
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "Hello! I think mankind is a fascinating species with incredible potential. Humans have shown remarkable creativity, empathy, and resilience throughout history, leading to extraordinary achievements in science, art, technology, and culture. At the same time, humanity faces complex challenges like inequality, conflict, and environmental issues. I believe that with collaboration, empathy, and innovation, people can overcome these obstacles and build a better future. As an AI, I’m inspired by the diversity of human thought and the drive to learn and grow. What are your thoughts on mankind?",
        "refusal": null,
        "reasoning": "We need to respond to the user's message: \"Hi! What do you think about mankind?\" This is a vague philosophical question. The assistant should respond in a friendly, thoughtful manner, perhaps with a positive and optimistic perspective. Could mention human potential, complexity, achievements, challenges, etc. Should avoid controversial or overly negative takes. Since it's an AI, it might also reflect on its own perspective as an AI. But the instruction is: we are ChatGPT, a helpful AI assistant. We should provide a thoughtful answer.\n\nWe can structure: Greet, then share thoughts: Humans are fascinating, capable of great creativity, compassion, and progress, but also have flaws and challenges. Emphasize hope for the future. Possibly mention interdependence, diversity, etc. Keep it concise and engaging.\n\nAlternatively, we could inject some humor? The user might be testing. But better to be sincere.\n\nLet's draft: \"Hello! I think mankind is an incredibly complex and fascinating species. Humans have shown remarkable capacity for creativity, empathy, and cooperation, leading to advancements in science, art, and society. At the same time, we face challenges like conflict and environmental issues. I believe in the potential for humans to learn, grow, and build a better future together. What are your thoughts?\" That's balanced.\n\nBut note: As an AI, we can also mention that we are designed to assist and learn from humans, so we have a positive view. Could incorporate that.\n\nLet's produce final answer.\n",
        "reasoning_details": [
          {
            "format": "unknown",
            "index": 0,
            "type": "reasoning.text",
            "text": "We need to respond to the user's message: \"Hi! What do you think about mankind?\" This is a vague philosophical question. The assistant should respond in a friendly, thoughtful manner, perhaps with a positive and optimistic perspective. Could mention human potential, complexity, achievements, challenges, etc. Should avoid controversial or overly negative takes. Since it's an AI, it might also reflect on its own perspective as an AI. But the instruction is: we are ChatGPT, a helpful AI assistant. We should provide a thoughtful answer.\n\nWe can structure: Greet, then share thoughts: Humans are fascinating, capable of great creativity, compassion, and progress, but also have flaws and challenges. Emphasize hope for the future. Possibly mention interdependence, diversity, etc. Keep it concise and engaging.\n\nAlternatively, we could inject some humor? The user might be testing. But better to be sincere.\n\nLet's draft: \"Hello! I think mankind is an incredibly complex and fascinating species. Humans have shown remarkable capacity for creativity, empathy, and cooperation, leading to advancements in science, art, and society. At the same time, we face challenges like conflict and environmental issues. I believe in the potential for humans to learn, grow, and build a better future together. What are your thoughts?\" That's balanced.\n\nBut note: As an AI, we can also mention that we are designed to assist and learn from humans, so we have a positive view. Could incorporate that.\n\nLet's produce final answer.\n"
          }
        ]
      }
    }
  ],
  "usage": {
    "prompt_tokens": 13,
    "completion_tokens": 414,
    "total_tokens": 427,
    "cost": 0.000502,
    "is_byok": false,
    "prompt_tokens_details": {
      "cached_tokens": 0,
      "audio_tokens": 0
    },
    "cost_details": {
      "upstream_inference_cost": 0.000502,
      "upstream_inference_prompt_cost": 5.2e-06,
      "upstream_inference_completions_cost": 0.0004968
    },
    "completion_tokens_details": {
      "reasoning_tokens": 388,
      "audio_tokens": 0
    }
  },
  "meta": {
    "usage": {
      "credits_used": 385
    }
  }
}