Text modelsAdvanced usersUpdated 2026-11-11

POST/v1/chat/completions

Kimi K3

Request ID k3

Protocols

POST /v1/chat/completions · Chat Completions example
POST /v1/messages · Anthropic Messages example
POST /v1/messages/count_tokens · Count Tokens example
POST /v1/responses · OpenAI Responses example
POST /v1/responses/compact · Responses Compact example

Authentication

Requests require an API key from API key management. For JSON requests, set Content-Type: application/json.

Authorization: Bearer YOUR_API_KEY
Anthropic Messages · x-api-key: YOUR_API_KEY

Request parameters

`POST /v1/chat/completions`

Field	Type	Required	Description
`model`	`string`	Required	Set the model ID to `k3`.
`messages`	`object[]`	Required	Send messages in conversation order. Each message has a `role` and `content`.
`messages[].content`	`string / object[]`	Conditional	The message body. Use plain text or an array of content blocks defined by the protocol.
`stream`	`boolean`	Optional	`true` requests an SSE stream. `false` requests a JSON response. Available modes depend on the endpoint.
`max_completion_tokens`	`integer`	Conditional	Caps generated tokens, including reasoning tokens where applicable. Use on endpoints that accept this field.
`max_tokens`	`integer`	Conditional	Compatibility field for the output token limit. Choose this field or `max_completion_tokens` according to the endpoint.
`temperature`	`number`	Conditional	Controls sampling randomness. Accepted values and whether the field takes effect depend on the model and endpoint.
`top_p`	`number`	Conditional	Limits sampling candidates by cumulative probability. Use when the model and endpoint support it.
`response_format`	`object`	Conditional	For structured output, use `type: json_object`, or use `type: json_schema` with `json_schema` when the endpoint supports schemas.
`reasoning_effort`	`string`	Conditional	Requested reasoning effort. Available levels and whether reasoning can be disabled depend on the model and endpoint.
`tools`	`object[]`	Conditional	Tool definitions available to the model. Tool types and parameter formats depend on the protocol and endpoint.
`tool_choice`	`string / object`	Conditional	Controls tool selection. Use a forced or named choice only when the selected endpoint supports it.

`POST /v1/messages`

Field	Type	Required	Description
`model`	`string`	Required	Set the model ID to `k3`.
`messages`	`object[]`	Required	Send `user` and `assistant` messages in order. Put system instructions in the top-level `system` field.
`messages[].content`	`string / object[]`	Required	The message body. Use plain text or an array of content blocks defined by the protocol.
`system`	`string / object[]`	Optional	System instructions as text or an array of text blocks.
`max_tokens`	`integer`	Required	Caps output tokens. The accepted limit depends on the model and endpoint.
`stream`	`boolean`	Optional	`true` requests an SSE stream. `false` requests a JSON response. Available modes depend on the endpoint.
`temperature`	`number`	Conditional	Controls sampling randomness. Accepted values and whether the field takes effect depend on the model and endpoint.
`top_p`	`number`	Conditional	Limits sampling candidates by cumulative probability. Use when the model and endpoint support it.
`tools`	`object[]`	Conditional	Tool definitions available to the model. Tool types and parameter formats depend on the protocol and endpoint.
`tool_choice`	`object`	Conditional	Controls tool selection. Use a forced or named choice only when the selected endpoint supports it.

`POST /v1/messages/count_tokens`

Field	Type	Required	Description
`model`	`string`	Required	Set the model ID to `k3`.
`messages`	`object[]`	Required	Send `user` and `assistant` messages in order. Put system instructions in the top-level `system` field.
`messages[].content`	`string / object[]`	Required	The message body. Use plain text or an array of content blocks defined by the protocol.
`system`	`string / object[]`	Optional	System instructions as text or an array of text blocks.
`tools`	`object[]`	Conditional	Tool definitions to include in the input token count. Send the same tools as in the Messages request.

`POST /v1/responses`

Field	Type	Required	Description
`model`	`string`	Required	Set the model ID to `k3`.
`input`	`string / object[]`	Required	Input text or an ordered array of input items. Message items use `role` and `content`.
`instructions`	`string`	Optional	Instructions describing the task and output requirements.
`max_output_tokens`	`integer`	Optional	Caps output tokens. The accepted limit depends on the model and endpoint.
`stream`	`boolean`	Optional	`true` requests an SSE stream. `false` requests a JSON response. Available modes depend on the endpoint.
`store`	`boolean`	Conditional	Where supported, `false` requests that the response not be retained for later retrieval.
`text.format`	`object`	Conditional	Set `text.format` for structured output, for example `type: json_schema`. Use only on endpoints that support the format.
`temperature`	`number`	Conditional	Controls sampling randomness. Accepted values and whether the field takes effect depend on the model and endpoint.
`top_p`	`number`	Conditional	Limits sampling candidates by cumulative probability. Use when the model and endpoint support it.
`reasoning.effort`	`string`	Conditional	Requested reasoning effort. Available levels and whether reasoning can be disabled depend on the model and endpoint.
`tools`	`object[]`	Conditional	Tool definitions available to the model. Tool types and parameter formats depend on the protocol and endpoint.
`tool_choice`	`string / object`	Conditional	Controls tool selection. Use a forced or named choice only when the selected endpoint supports it.

`POST /v1/responses/compact`

Field	Type	Required	Description
`model`	`string`	Required	Set the model ID to `k3`.
`input`	`string / object[]`	Required	Context to compact. Send text or the conversation input items in their original order. Accepted formats depend on the endpoint.

Request examples

Set your API key in the local ICODEEASY_API_KEY environment variable and choose an endpoint that accepts direct HTTP calls. For endpoints that require a specific client, follow that client’s setup guide.

Chat Completions

bash

curl -sS 'https://api.icodeeasy.cc/v1/chat/completions' \
  -H "Authorization: Bearer $ICODEEASY_API_KEY" \
  -H 'Content-Type: application/json' \
  --data '{
  "model": "k3",
  "messages": [
    {
      "role": "user",
      "content": "Explain how a rainbow forms in one sentence."
    }
  ],
  "stream": false
}'

Anthropic Messages

bash

curl -sS 'https://api.icodeeasy.cc/v1/messages' \
  -H "x-api-key: $ICODEEASY_API_KEY" \
  -H 'Content-Type: application/json' \
  -H 'anthropic-version: 2023-06-01' \
  --data '{
  "model": "k3",
  "max_tokens": 256,
  "messages": [
    {
      "role": "user",
      "content": "Explain how a rainbow forms in one sentence."
    }
  ]
}'

Count Tokens

bash

curl -sS 'https://api.icodeeasy.cc/v1/messages/count_tokens' \
  -H "x-api-key: $ICODEEASY_API_KEY" \
  -H 'Content-Type: application/json' \
  -H 'anthropic-version: 2023-06-01' \
  --data '{
  "model": "k3",
  "messages": [
    {
      "role": "user",
      "content": "Explain how a rainbow forms in one sentence."
    }
  ]
}'

OpenAI Responses

bash

curl -N -sS 'https://api.icodeeasy.cc/v1/responses' \
  -H "Authorization: Bearer $ICODEEASY_API_KEY" \
  -H 'Content-Type: application/json' \
  --data '{
  "model": "k3",
  "instructions": "You are a helpful assistant.",
  "input": [
    {
      "type": "message",
      "role": "user",
      "content": [
        {
          "type": "input_text",
          "text": "Explain how a rainbow forms in one sentence."
        }
      ]
    }
  ],
  "stream": true,
  "store": false
}'

Responses Compact

bash

curl -sS 'https://api.icodeeasy.cc/v1/responses/compact' \
  -H "Authorization: Bearer $ICODEEASY_API_KEY" \
  -H 'Content-Type: application/json' \
  --data '{
  "model": "k3",
  "input": "Explain how a rainbow forms in one sentence."
}'

Read the response

Chat Completions · Read text from choices[].message.content. In an SSE stream, incremental output appears in choices[].delta. JSON · SSE
Anthropic Messages · Read blocks of type text in content[]. In a stream, process SSE content block events in order. JSON · SSE
Count Tokens · Read the input token count from input_tokens.
OpenAI Responses · In a stream, append the delta fields of response.output_text.delta events and handle completion and failure events. SSE

Full integration guide

Capabilities and limits

streaming · Supported
tool_calling · Supported
image_input · Not published
video_input · Not published

Compatibility aliases and retired redirects

kimi-k3, k3[1m] → k3 (Alias)

Model Capability Catalog · View pricing

Kimi Code Setup: API Key, K3.1 Pro, and Claude Code

Set up Kimi Code CLI or Claude Code with K3.1 Pro. Compare persistent and temporary settings, verify your first call, and fix 401, model, and URL errors.

Chat Completions

Send a conversation using the OpenAI Chat format. Check that your model and selected endpoint support the Chat protocol before sending the request.

Anthropic Messages

Send a request in the Anthropic Messages format. This page covers direct HTTP calls, with Claude Code setup covered in the related guides.

Kimi K3

Protocols

Authentication

Request parameters

POST /v1/chat/completions

POST /v1/messages

POST /v1/messages/count_tokens

POST /v1/responses

POST /v1/responses/compact

Request examples

Chat Completions

Anthropic Messages

Count Tokens

OpenAI Responses

Responses Compact

Read the response

Capabilities and limits

Compatibility aliases and retired redirects

Related guides

Kimi Code Setup: API Key, K3.1 Pro, and Claude Code

Chat Completions

Anthropic Messages

`POST /v1/chat/completions`

`POST /v1/messages`

`POST /v1/messages/count_tokens`

`POST /v1/responses`

`POST /v1/responses/compact`