MiMo - Xiaomi

Updated at:

This topic describes how to call Xiaomi's MiMo series models on Alibaba Cloud Model Studio.

Importantxiaomi/mimo-v2.5-pro, supplied by Xiaomi, will be discontinued on October 21, 2026. Switch to xiaomi/mimo-v2.6-pro-ultraspeed.

ImportantThis document applies only to the China (Beijing) region. To use the models, you must obtain an API key from the China (Beijing) region.

Quick start

mimo-v2.5-pro is a hybrid reasoning model provided by Xiaomi. It runs in thinking mode by default (enable_thinking defaults to true). To disable thinking and get direct responses, explicitly pass enable_thinking: false. The following example calls mimo-v2.5-pro in thinking mode with streaming enabled.

Before you begin, make sure you have Obtain an API key and Configure API key as an environment variable. If you are using an SDK, you also need to Install the SDK.

OpenAI-compatible

Noteenable_thinking is not a standard OpenAI parameter. In the OpenAI Python SDK, pass it via extra_body. In the Node.js SDK, pass it as a top-level parameter.

Python

Sample code

from openai import OpenAI
import os

client = OpenAI(
    # Replace with your Model Studio API Key: api_key="sk-xxx"
    api_key=os.getenv("DASHSCOPE_API_KEY"),
    # China (Beijing) region. Replace {WorkspaceId} with your Bailian workspace ID. URLs vary by region.
    base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1",
)

messages = [{"role": "user", "content": "What is 1+1?"}]
completion = client.chat.completions.create(
    model="xiaomi/mimo-v2.6-pro-ultraspeed",
    messages=messages,
    # Thinking mode is enabled by default. To disable: extra_body={"enable_thinking": False}
    stream=True,
    stream_options={
        "include_usage": True
    },
)

reasoning_content = ""  # Full thinking process
answer_content = ""  # Final answer
is_answering = False

print("\n" + "=" * 20 + " Thinking " + "=" * 20 + "\n")

for chunk in completion:
    if not chunk.choices:
        print("\n" + "=" * 20 + " Usage " + "=" * 20 + "\n")
        print(chunk.usage)
        continue

    delta = chunk.choices[0].delta

    if hasattr(delta, "reasoning_content") and delta.reasoning_content is not None:
        if not is_answering:
            print(delta.reasoning_content, end="", flush=True)
        reasoning_content += delta.reasoning_content

    if hasattr(delta, "content") and delta.content:
        if not is_answering:
            print("\n" + "=" * 20 + " Answer " + "=" * 20 + "\n")
            is_answering = True
        print(delta.content, end="", flush=True)
        answer_content += delta.content

Sample response

==================== Thinking ====================

The user asked a simple arithmetic question: what is 1+1.
This is a basic addition operation. The answer is 2. I'll give a direct answer.
==================== Answer ====================

1+1 equals 2. This is a basic addition operation.
==================== Usage ====================

CompletionUsage(completion_tokens=42, prompt_tokens=9, total_tokens=51, prompt_tokens_details={'cached_tokens': 0})

Node.js

Sample code

import OpenAI from "openai";
import process from 'process';

const openai = new OpenAI({
    // Replace with your Model Studio API Key: apiKey: "sk-xxx"
    apiKey: process.env.DASHSCOPE_API_KEY,
    // China (Beijing) region. Replace {WorkspaceId} with your Bailian workspace ID. URLs vary by region.
    baseURL: 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'
});

let reasoningContent = '';
let answerContent = '';
let isAnswering = false;

async function main() {
    try {
        const messages = [{ role: 'user', content: 'What is 1+1?' }];

        const stream = await openai.chat.completions.create({
            model: 'xiaomi/mimo-v2.6-pro-ultraspeed',
            messages,
            // Thinking mode is enabled by default. To disable: enable_thinking: false
            stream: true,
            stream_options: {
                include_usage: true
            },
        });

        console.log('\n' + '='.repeat(20) + ' Thinking ' + '='.repeat(20) + '\n');

        for await (const chunk of stream) {
            if (!chunk.choices?.length) {
                console.log('\n' + '='.repeat(20) + ' Usage ' + '='.repeat(20) + '\n');
                console.log(chunk.usage);
                continue;
            }

            const delta = chunk.choices[0].delta;

            if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {
                if (!isAnswering) {
                    process.stdout.write(delta.reasoning_content);
                }
                reasoningContent += delta.reasoning_content;
            }

            if (delta.content !== undefined && delta.content) {
                if (!isAnswering) {
                    console.log('\n' + '='.repeat(20) + ' Answer ' + '='.repeat(20) + '\n');
                    isAnswering = true;
                }
                process.stdout.write(delta.content);
                answerContent += delta.content;
            }
        }
    } catch (error) {
        console.error('Error:', error);
    }
}

main();

Sample response

==================== Thinking ====================

The user asked a simple arithmetic question: what is 1+1.
This is a basic addition operation. The answer is 2. I'll give a direct answer.
==================== Answer ====================

1+1 equals 2. This is a basic addition operation.
==================== Usage ====================

{ prompt_tokens: 9, completion_tokens: 42, total_tokens: 51, prompt_tokens_details: { cached_tokens: 0 } }

HTTP

Sample code

curl

# China (Beijing) region. Replace {WorkspaceId} with your Bailian workspace ID. URLs vary by region.
curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \
-H "Authorization: Bearer $DASHSCOPE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
    "model": "xiaomi/mimo-v2.6-pro-ultraspeed",
    "messages": [
        {
            "role": "user",
            "content": "What is 1+1?"
        }
    ],
    "stream": true,
    "stream_options": {
        "include_usage": true
    }
}'

Multimodal input

xiaomi/mimo-v2.6-pro-ultraspeed supports image, audio, and video input. Pass text and multimodal content together in the content array of the user message: use image_url for images, input_audio for audio with format specifying the audio format, and video_url for videos.

Image

# The following URL is for the China (Beijing) region. Replace {WorkspaceId} with your Model Studio workspace ID. URLs vary by region.
curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \
-H "Authorization: Bearer $DASHSCOPE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
    "model": "xiaomi/mimo-v2.6-pro-ultraspeed",
    "messages": [
        {
            "role": "user",
            "content": [
                {"type": "text", "text": "What scene is depicted in the image?"},
                {"type": "image_url", "image_url": {"url": "https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg"}}
            ]
        }
    ]
}'

Audio

# The following URL is for the China (Beijing) region. Replace {WorkspaceId} with your Model Studio workspace ID. URLs vary by region.
curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \
-H "Authorization: Bearer $DASHSCOPE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
    "model": "xiaomi/mimo-v2.6-pro-ultraspeed",
    "messages": [
        {
            "role": "user",
            "content": [
                {"type": "text", "text": "What is said in the audio?"},
                {"type": "input_audio", "input_audio": {"data": "http://paas-aliyun-beijing.oss-cn-beijing.aliyuncs.com/qwen-translation/Cindy.wav", "format": "wav"}}
            ]
        }
    ]
}'

Video

# The following URL is for the China (Beijing) region. Replace {WorkspaceId} with your Model Studio workspace ID. URLs vary by region.
curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \
-H "Authorization: Bearer $DASHSCOPE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
    "model": "xiaomi/mimo-v2.6-pro-ultraspeed",
    "messages": [
        {
            "role": "user",
            "content": [
                {"type": "text", "text": "What is in the video?"},
                {"type": "video_url", "video_url": {"url": "http://paas-aliyun-beijing.oss-cn-beijing.aliyuncs.com/qwen-translation/normal_video.mp4"}}
            ]
        }
    ]
}'

Supported features

Feature

Supported

Notes

Multi-turn conversation

Supported

In thinking mode, you must include the reasoning_content field in each assistant message, otherwise the API returns an error

Function calling

Supported

tool_choice only supports auto. Tool function names may only contain a-z, A-Z, 0-9, underscores, and hyphens, with a maximum length of 64

Context caching

Supported

Implicit caching, automatically enabled

Structured output

Supported

response_format parameter does not support json_schema

Web search

Not supported

—

Prefix completion

Not supported

—

Multimodal input

Model-dependent

xiaomi/mimo-v2.6-pro-ultraspeed supports image, audio, and video input. xiaomi/mimo-v2.5-pro does not, and passing image, audio, or video input to it returns an error

mimo-v2.5-pro does not support the following parameters: top_k, reasoning_effort, thinking_budget, modalities, repetition_penalty, vl_high_resolution_images, preserve_thinking, tool_stream, enable_code_interpreter, parallel_tool_calls, seed, logprobs, top_logprobs, n,audio,enable_search,search_options,X-DashScope-DataInspection,skill. These parameters are ignored or return an error if passed.

Some parameter ranges and functional behaviors in MiMo differ from those in the Bailian platform. Please refer to the table below for details:

Full screen copy

Parameter

ModelStudio

MiMo

temperature

Range: [0, 2)

mimo-v2.5-pro: Range: [0, 1.5], Default: 1.0

mimo-v2.6-pro-ultraspeed: Not customizable. If you pass another value, 1.0 still takes effect

top_p

Range: (0, 1.0]

mimo-v2.5-pro: Range: [0.01, 1.0], Default: 0.95

mimo-v2.6-pro-ultraspeed: Not customizable. If you pass another value, 0.95 still takes effect

presence_penalty

—

Range: [-2, 2], Default: 0

stop

—

Up to 4 sequences

max_tokens

Limits output length only (does not limit CoT length)

Limits both output and Chain-of-Thought (CoT) length

Models and pricing

The MiMo series models are hybrid reasoning models provided by Xiaomi, supporting both thinking and non-thinking modes.

For context length and pricing details, refer to the Model Studio console.

Billing is based on input and output tokens.

In thinking mode, the chain-of-thought is billed as output tokens.

Error codes

The MiMo series models are provided directly by Xiaomi. Their error codes differ from the standard Model Studio error codes. When calling MiMo models, refer to the table below.

Error code

Cause

Solution

400

Invalid request body

  • Verify the JSON format is correct

  • Verify all required parameters are included

  • Verify parameter values are within valid ranges

  • Verify the message format meets API requirements

  • Verify field names are correct

  • In multi-turn thinking mode, you must pass back the complete reasoning_content field

401

Missing or invalid API Key, or incorrect Authorization header format

Verify your API Key and Authorization header format are correct

402

Insufficient account balance

Check your account balance and top up as needed

403

Service not available in the current region, or API Key is blocked

Create a new API Key and ensure your input content complies with safety policies

404

The endpoint or model does not support image input

Confirm that the model or endpoint supports multimodal image input

421

Content moderation triggered

Avoid inputting unsafe or sensitive content

429

Too many requests, or Token Plan quota exhausted

  • Implement exponential backoff and retry logic, or reduce request frequency

  • Upgrade your Token Plan or switch to pay-as-you-go

500

Internal server error

Retry later, or contact us for assistance

503

Server overloaded

Retry later