API

Every deepseek.api.* operation is listed below with parameter shapes and return types from the plugin Zod schemas.

New to Corsair? See API access, authentication, and error handling.

Anthropic

createMessage

anthropic.createMessage Generate a response via DeepSeek’s Anthropic-compatible Messages API, with support for system prompts, tool calling, and thinking mode Risk: write

await corsair.deepseek.api.anthropic.createMessage({});

Input

Name	Type	Required	Description
`model`	`deepseek-chat \| deepseek-reasoner`	Yes	—
`maxTokens`	`number`	Yes	—
`messages`	`object[]`	Yes	—
`system`	`object[]`	No	—
`stopSequences`	`string[]`	No	—
`temperature`	`number`	No	—
`topP`	`number`	No	—
`topK`	`number`	No	—
`tools`	`object[]`	No	—
`toolChoice`	`object`	No	—
`thinking`	`object`	No	—
`metadata`	`object`	No	—

messages full type

{
  role: user | assistant,
  content: string | (
    {
      type: text,
      text: string
    } | {
      type: tool_use,
      id: string,
      name: string,
      input: {
      }
    } | {
      type: tool_result,
      tool_use_id: string,
      content?: string | {
        type: text,
        text: string
      }[],
      is_error?: boolean
    } | {
      type: thinking,
      thinking: string,
      signature: string
    } | {
      type: redacted_thinking,
      data: string
    }
  )[]
}[]

system full type

string | {
  type: text,
  text: string
}[]

tools full type

{
  name: string,
  description?: string,
  input_schema: {
  }
}[]

toolChoice full type

{
  type: auto,
  disable_parallel_tool_use?: boolean
} | {
  type: any,
  disable_parallel_tool_use?: boolean
} | {
  type: tool,
  name: string,
  disable_parallel_tool_use?: boolean
}

thinking full type

{
  type: enabled,
  budget_tokens: number
} | {
  type: disabled
}

metadata full type

{
  user_id?: string
}

Output

Name	Type	Required	Description
`id`	`string`	Yes	—
`type`	`message`	Yes	—
`role`	`assistant`	Yes	—
`model`	`string`	Yes	—
`content`	`object[]`	Yes	—
`stop_reason`	`end_turn \| max_tokens \| stop_sequence \| tool_use`	No	—
`stop_sequence`	`string`	No	—
`usage`	`object`	Yes	—

content full type

(
  {
    type: text,
    text: string
  } | {
    type: tool_use,
    id: string,
    name: string,
    input: {
    }
  } | {
    type: tool_result,
    tool_use_id: string,
    content?: string | {
      type: text,
      text: string
    }[],
    is_error?: boolean
  } | {
    type: thinking,
    thinking: string,
    signature: string
  } | {
    type: redacted_thinking,
    data: string
  }
)[]

usage full type

{
  input_tokens: number,
  output_tokens: number,
  cache_creation_input_tokens?: number | null,
  cache_read_input_tokens?: number | null
}

Chat

createCompletion

chat.createCompletion Generate an AI chat response using deepseek-chat or deepseek-reasoner, with support for temperature, tool calling, and structured output Risk: write

await corsair.deepseek.api.chat.createCompletion({});

Input

Name	Type	Required	Description
`model`	`deepseek-chat \| deepseek-reasoner`	Yes	—
`messages`	`object[]`	Yes	—
`frequencyPenalty`	`number`	No	—
`maxTokens`	`number`	No	—
`presencePenalty`	`number`	No	—
`responseFormat`	`object`	No	—
`stop`	`string \| string[]`	No	—
`temperature`	`number`	No	—
`topP`	`number`	No	—
`tools`	`object[]`	No	—
`toolChoice`	`object`	No	—
`logprobs`	`boolean`	No	—
`topLogprobs`	`number`	No	—

messages full type

{
  role: system | user | assistant | tool,
  content?: string | null,
  name?: string,
  tool_calls?: {
    id: string,
    type: function,
    function: {
      name: string,
      arguments: string
    }
  }[],
  tool_call_id?: string,
  prefix?: boolean
}[]

responseFormat full type

{
  type: text
} | {
  type: json_object
}

tools full type

{
  type: function,
  function: {
    name: string,
    description?: string,
    parameters?: {
    }
  }
}[]

toolChoice full type

none | auto | required | {
  type: function,
  function: {
    name: string
  }
}

Output

Name	Type	Required	Description
`id`	`string`	Yes	—
`object`	`chat.completion`	Yes	—
`created`	`number`	Yes	—
`model`	`string`	Yes	—
`choices`	`object[]`	Yes	—
`system_fingerprint`	`string`	No	—
`usage`	`object`	No	—

choices full type

{
  index: number,
  message: {
    role: assistant,
    content?: string | null,
    reasoning_content?: string | null,
    tool_calls?: {
      id: string,
      type: function,
      function: {
        name: string,
        arguments: string
      }
    }[]
  },
  logprobs?: {
    content?: any[] | null
  } | null,
  finish_reason: stop | length | content_filter | tool_calls | insufficient_system_resource
}[]

usage full type

{
  prompt_tokens: number,
  completion_tokens: number,
  total_tokens: number,
  prompt_cache_hit_tokens?: number,
  prompt_cache_miss_tokens?: number,
  completion_tokens_details?: {
    reasoning_tokens?: number
  }
}

Models

list

models.list List the DeepSeek models currently available to the account Risk: read

await corsair.deepseek.api.models.list({});

Input: empty object Output

Name	Type	Required	Description
`object`	`list`	Yes	—
`data`	`object[]`	Yes	—

data full type

{
  id: string,
  object: model,
  owned_by: string
}[]

User

getBalance

user.getBalance Get the current account balance, including granted and topped-up amounts broken down by currency Risk: read

await corsair.deepseek.api.user.getBalance({});

Input: empty object Output

Name	Type	Required	Description
`is_available`	`boolean`	Yes	—
`balance_infos`	`object[]`	Yes	—

balance_infos full type

{
  currency: CNY | USD,
  total_balance: string,
  granted_balance: string,
  topped_up_balance: string
}[]

Overview

Plugins

Anthropic

createMessage

Chat

createCompletion

Models

list

User

getBalance

​Anthropic

​createMessage

​Chat

​createCompletion

​Models

​list

​User

​getBalance

Anthropic

createMessage

Chat

createCompletion

Models

list

User

getBalance