Skip to main content
POST

Authorizations

​
Authorization
string
header
required

Pass Authorization: Bearer <token> in the request header.

Body

application/json

MIMO chat completions request body.

​
model
enum<string>
required

Model ID to use

Available options:
mimo-v2.5
Example:

"mimo-v2.5"

​
messages
(Developer message · object · object | System message · object · object | User message · object · object | Assistant message · object · object | Tool message · object · object)[]
required

A list of messages comprising the conversation so far.

Minimum array length: 1

A single message in the conversation. Depending on the role, it is divided into five message types: developer, system, user, assistant, and tool.

​
frequency_penalty
number | null
default:0

A value between -2.0 and 2.0. Positive values penalize new tokens based on their existing frequency in the text so far, reducing the model's likelihood of repeating the same content.

Required range: -2 <= x <= 2
​
presence_penalty
number | null
default:0

Number between -2.0 and 2.0. Positive values penalize new tokens based on whether they have appeared in the text so far, increasing the model's likelihood to talk about new topics.

Required range: -2 <= x <= 2
​
max_completion_tokens
integer | null

The upper limit of tokens that can be generated in the chat completion, including visible output tokens and reasoning tokens. Default is 32768 for mimo-v2.5. Range: [1, 131072]

​
response_format
Text · object

Default response format. Used for generating text responses.

​
stop

Up to 4 sequences where the API will stop generating further tokens when any of them is encountered. The returned text will not contain the stop sequences.

​
stream
boolean | null
default:false

If set to true, the model's response data will be streamed to the client via SSE (server-sent events) during generation.

​
thinking
object

Controls whether the model enables chain-of-thought (thinking) mode. Note: In multi-round tool calls under thinking mode, the model returns the reasoning_content field along with tool_calls. To continue the conversation, it is recommended to retain all historical reasoning_content in the messages array of each subsequent request for optimal performance. In thinking mode, mimo-v2.5-pro and mimo-v2.5 do not support custom temperature and top_p parameters. Even if passed, the model will force the recommended default values of 1.0 and 0.95 respectively.

​
temperature
number
default:1

Sampling temperature to use, between 0 and 1.5. Higher values (e.g., 0.8) make the output more random, while lower values (e.g., 0.2) make it more focused and deterministic. We generally recommend altering this or top_p but not both. In thinking mode, mimo-v2.5-pro and mimo-v2.5 do not support custom temperature. Even if passed, the model will force the recommended default value of 1.0.

Required range: 0 <= x <= 1.5
​
tool_choice
enum<string>

Controls how the model selects tools. Note: When tool_choice is set to a non-auto value, the backend will remove this field by default, and the model's response behavior will still be equivalent to auto mode (this logic may be subject to adjustment). Allowed values: auto

Available options:
auto
​
top_p
number
default:0.95

Probability threshold for nucleus sampling, controlling the diversity of text generation. Higher top_p values produce more diverse text; lower values produce more deterministic text. Since both temperature and top_p control text diversity, it is recommended to adjust only one of them. In thinking mode, mimo-v2.5-pro and mimo-v2.5 do not support custom top_p. Even if passed, the model will force the recommended default value of 0.95.

Required range: 0.01 <= x <= 1
​
tools
(Function tool · object | Web search tool · object)[]

A list of tools the model may call. Currently, both function tools and web_search tools are supported. Note: In multi-round tool calls under thinking mode, the model returns the reasoning_content field along with tool_calls. To continue the conversation, it is recommended to retain all historical reasoning_content in the messages array of each subsequent request for optimal performance.

Function tool definition.

Response

Success. Returns JSON in non-streaming mode; returns an SSE event stream in streaming mode.

Chat response object (non-streaming output).

​
id
string

A unique identifier for the response.

​
object
string

The type of the object, which is chat.completion.

​
created
integer

The Unix timestamp (in seconds) when the chat completion was created.

​
model
string

The name of the model that generated the completion.

​
choices
object[]

A list of completions generated by the model.

​
usage
object

Usage information for this chat completion request.