Deepseek

Configure Deepseek specification.

When using Deepseek LLMs, you have the choice of using a built-in Graphlit model, which accrues credits for usage, or using your own Deepseek API key.

Graphlit currently supports these built-in Deepseek models via the model field: DEEPSEEK_CHAT, DEEPSEEK_CODER.

By assigning the model field to CUSTOM, you also will need to assign the key and modelName to use your own Deepseek developer account.

You can find the list of current Deepseek models here, and use the model name syntax such as deepseek-chat.

Create Built-In Model Specification

Mutation:

mutation CreateSpecification($specification: SpecificationInput!) {
  createSpecification(specification: $specification) {
    id
    name
    state
    type
    serviceType
  }
}

Variables:

{
  "specification": {
    "type": "COMPLETION",
    "serviceType": "DEEPSEEK",
    "deepseek": {
      "model": "DEEPSEEK_CHAT"
    },
    "name": "Deepseek"
  }
}

Response:

{
  "type": "COMPLETION",
  "serviceType": "DEEPSEEK",
  "id": "a87ce6cd-75f4-4391-8d00-33149e9afc86",
  "name": "Deepseek",
  "state": "ENABLED"
}

Create Custom Model Specification

Mutation:

mutation CreateSpecification($specification: SpecificationInput!) {
  createSpecification(specification: $specification) {
    id
    name
    state
    type
    serviceType
  }
}

Variables:

{
  "specification": {
    "type": "COMPLETION",
    "serviceType": "DEEPSEEK",
    "deepseek": {
      "key": "redacted",
      "modelName": "deepseek-chat",
      "model": "CUSTOM"
    },
    "name": "Deepseek Custom"
  }
}

Response:

{
  "type": "COMPLETION",
  "serviceType": "DEEPSEEK",
  "id": "3c7d107c-cc9f-4c2f-8ac5-0b0b7ddd2ddf",
  "name": "Deepseek Custom",
  "state": "ENABLED"
}