Create a client model

curl --request POST \
  --url https://api.agents.slng.ai/v1/agents/client-models \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "service_type": "llm",
  "provider": "openai-compat",
  "url": "https://api.openai.com/v1",
  "api_key": "sk-proj-example",
  "model_name": "gpt-4o-mini",
  "model_id": "gpt-4o-mini"
}
'

{
  "id": "6a3f5b2c-1d4e-4f8a-9b0c-2e3d4f5a6b7c",
  "org_id": "1b2c3d4e-5f6a-7b8c-9d0e-1f2a3b4c5d6e",
  "api_key_id": null,
  "service_type": "llm",
  "provider": "openai-compat",
  "model_lab": null,
  "url": "https://api.openai.com/v1",
  "has_api_key": true,
  "auth_header": null,
  "model_name": "gpt-4o-mini",
  "model_id": null,
  "azure_deployment": null,
  "api_version": null,
  "has_vertex_credentials": false,
  "has_aws_credentials": false,
  "vertex_location": null,
  "vertex_project": null,
  "languages": [],
  "regions": null,
  "enabled": true,
  "kwargs": null,
  "created_at": "2026-01-15T09:30:00Z",
  "updated_at": "2026-01-15T09:30:00Z"
}

{
  "detail": "client model not found",
  "error": {
    "code": "NOT_FOUND",
    "message": "client model not found",
    "request_id": "req_01H8XYZ",
    "retryable": false
  }
}

{
  "detail": "client model not found",
  "error": {
    "code": "NOT_FOUND",
    "message": "client model not found",
    "request_id": "req_01H8XYZ",
    "retryable": false
  }
}

{
  "detail": "client model not found",
  "error": {
    "code": "NOT_FOUND",
    "message": "client model not found",
    "request_id": "req_01H8XYZ",
    "retryable": false
  }
}

{
  "detail": "client model not found",
  "error": {
    "code": "NOT_FOUND",
    "message": "client model not found",
    "request_id": "req_01H8XYZ",
    "retryable": false
  }
}

{
  "detail": "client model not found",
  "error": {
    "code": "NOT_FOUND",
    "message": "client model not found",
    "request_id": "req_01H8XYZ",
    "retryable": false
  }
}

{
  "detail": "client model not found",
  "error": {
    "code": "NOT_FOUND",
    "message": "client model not found",
    "request_id": "req_01H8XYZ",
    "retryable": false
  }
}

POST

agents

client-models

curl --request POST \
  --url https://api.agents.slng.ai/v1/agents/client-models \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "service_type": "llm",
  "provider": "openai-compat",
  "url": "https://api.openai.com/v1",
  "api_key": "sk-proj-example",
  "model_name": "gpt-4o-mini",
  "model_id": "gpt-4o-mini"
}
'

{
  "id": "6a3f5b2c-1d4e-4f8a-9b0c-2e3d4f5a6b7c",
  "org_id": "1b2c3d4e-5f6a-7b8c-9d0e-1f2a3b4c5d6e",
  "api_key_id": null,
  "service_type": "llm",
  "provider": "openai-compat",
  "model_lab": null,
  "url": "https://api.openai.com/v1",
  "has_api_key": true,
  "auth_header": null,
  "model_name": "gpt-4o-mini",
  "model_id": null,
  "azure_deployment": null,
  "api_version": null,
  "has_vertex_credentials": false,
  "has_aws_credentials": false,
  "vertex_location": null,
  "vertex_project": null,
  "languages": [],
  "regions": null,
  "enabled": true,
  "kwargs": null,
  "created_at": "2026-01-15T09:30:00Z",
  "updated_at": "2026-01-15T09:30:00Z"
}

{
  "detail": "client model not found",
  "error": {
    "code": "NOT_FOUND",
    "message": "client model not found",
    "request_id": "req_01H8XYZ",
    "retryable": false
  }
}

{
  "detail": "client model not found",
  "error": {
    "code": "NOT_FOUND",
    "message": "client model not found",
    "request_id": "req_01H8XYZ",
    "retryable": false
  }
}

{
  "detail": "client model not found",
  "error": {
    "code": "NOT_FOUND",
    "message": "client model not found",
    "request_id": "req_01H8XYZ",
    "retryable": false
  }
}

{
  "detail": "client model not found",
  "error": {
    "code": "NOT_FOUND",
    "message": "client model not found",
    "request_id": "req_01H8XYZ",
    "retryable": false
  }
}

{
  "detail": "client model not found",
  "error": {
    "code": "NOT_FOUND",
    "message": "client model not found",
    "request_id": "req_01H8XYZ",
    "retryable": false
  }
}

{
  "detail": "client model not found",
  "error": {
    "code": "NOT_FOUND",
    "message": "client model not found",
    "request_id": "req_01H8XYZ",
    "retryable": false
  }
}

Authorizations

Authorization

string

header

required

Your SLNG consumer API key.

Body

application/json

The client model to create, including its write-only provider credentials.

The writable fields of a client model (a bring-your-own model provider credential). The organisation is taken from the authenticated request. The api_key, vertex_credentials, and aws_credentials fields are write-only and are never returned.

model_name

string

required

Your alias for the model. Sent upstream as-is unless model_id is set.

Required string length: 1 - 255

service_type

enum<string>

default:llm

The kind of model: llm for text generation, stt for speech-to-text, or tts for text-to-speech.

Available options:

llm,

stt,

tts

url

string | null

Required string length: 1 - 2000

api_key

string | null

The upstream provider API key. Write-only: stored encrypted and never returned. Required for openai-compat, openai-responses, and azure; rejected for vertex.

Minimum string length: 1

model_id

string | null

The upstream provider-facing model id when it differs from model_name. Empty or null sends model_name as-is.

Maximum string length: 255

provider

enum<string> | null

The bring-your-own model provider. openai-compat targets an OpenAI-compatible Chat Completions endpoint; openai-responses targets the OpenAI /v1/responses endpoint with the same config as openai-compat; azure is Azure OpenAI; vertex is Google Vertex AI.

Available options:

openai-compat,

openai-responses,

azure,

vertex

model_lab

string | null

Maximum string length: 100

azure_deployment

string | null

Azure OpenAI deployment name. Azure only.

Maximum string length: 255

api_version

string | null

Azure OpenAI API version. Azure only.

Maximum string length: 100

auth_header

string | null

openai-compat and openai-responses only: a custom credential header name that replaces the default Authorization Bearer scheme. Rejected for azure and vertex.

Maximum string length: 64

vertex_credentials

Vertex Credentials · object | null

Vertex only: the GCP service-account key JSON. Write-only; stored encrypted and never returned (responses expose has_vertex_credentials).

vertex_location

string | null

Maximum string length: 100

vertex_project

string | null

Maximum string length: 100

aws_credentials

Aws Credentials · object | null

Speech BYOK only: AWS SigV4 credentials (access_key_id, secret_access_key, region). Write-only; stored encrypted and never returned (responses expose has_aws_credentials).

api_key_id

string<uuid> | null

Speech BYOK only: the API key id this credential is scoped to. Null means the organisation default. Rejected on llm rows. Non-secret.

languages

string[]

regions

string[] | null

enabled

boolean

default:true

kwargs

Kwargs · object | null

fallbacks

any[] | null

deprecated

Retired. Leave it out, or send null or an empty list; a non-empty list returns a 422. To choose which key answers first, use Set as default on the BYOK page of the Vault (organization owners and admins). To add fallbacks for an agent, use Fallback routing in its Think section.

Response

The created client model.

A client model as returned by the API. Credential values are never included; has_api_key, has_vertex_credentials, and has_aws_credentials report whether each secret is stored.

string<uuid>

required

org_id

string<uuid>

required

service_type

enum<string>

required

The kind of model: llm, stt, or tts.

Available options:

llm,

stt,

tts

provider

string | null

required

The configured provider. May be a legacy free-form label on records created before the current provider set.

model_lab

string | null

required

url

string | null

required

has_api_key

boolean

required

Whether an upstream API key is stored. The key value is write-only and never returned.

auth_header

string | null

required

The custom credential header name, or null for the default Authorization Bearer scheme. The credential value itself is never returned.

model_name

string

required

The configured alias for the model.

model_id

string | null

required

The upstream provider-facing model id, or null when model_name is sent as-is.

azure_deployment

string | null

required

The configured Azure OpenAI deployment name (Azure only).

api_version

string | null

required

The configured Azure OpenAI API version (Azure only).

has_vertex_credentials

boolean

required

Whether Vertex service-account credentials are stored. The value is write-only and never returned.

has_aws_credentials

boolean

required

Whether AWS SigV4 credentials are stored. The value is write-only and never returned.

vertex_location

string | null

required

vertex_project

string | null

required

languages

string[]

required

regions

string[] | null

required

enabled

boolean

required

kwargs

Kwargs · object | null

required

created_at

string<date-time>

required

updated_at

string<date-time>

required

api_key_id

string<uuid> | null

Speech BYOK only: the API key id this credential is scoped to, or null for the organisation default. Non-secret.

Speech to text

Context Router (LLM)

Text to speech

Agents

Unified API

Metadata API

Authorizations

Body

Response