Elastic Deployment

Elastic Deployment is a flexible orchestration feature provided by the Yotta SaaS platform that enables users to quickly create, scale, and manage GPU-powered workloads based on custom images.

API List

Create Elastic Deployment

Endpoint: /openapi/v1/elastic/deploy/create

Method: POST

Description: Creates a new Elastic Deployment on the platform.

Authorizations

Name	Location	Type	Required	Description
X-API-Key	header	string	required	Your API Key

Request Body (JSON)

Name	Type	Required	Description
name	string	required	Elastic Deployment name
image	string	required	Docker Image
imageRegistry	string	optional	Image registry URL; defaults to Docker Hub. Example: https://registry.example.com
resources	Resource[]	required	GPU-related resources. Workers will be deployed on instances satisfying the specified resources. Multiple resource configurations can be specified. Note: Each Resource must specify the same gpuCount.
minSingleCardVramInGb	number	optional	Minimum VRAM required per single card (GB)
minSingleCardVcpu	number	optional	Minimum VCPU required per single card
minSingleCardRamInGb	number	optional	Minimum RAM required per single card (GB)
containerVolumeInGb	number	required	Container ephemeral volume size (GB)
credentialId	string	optional	Required if using a private image. To create a credential, refer to [Credential Module -> Create Credential].
workers	number	required	Number of workers to deploy
serviceMode	string	required	Supports three Service Modes: ALB: Provides service via HTTP requests. QUEUE: Tasks are submitted to the Yotta Queue service for asynchronous processing. CUSTOM: Neither HTTP requests nor the Queue service are required.
initializationCommand	string	required	Initialization command executed upon worker startup
environmentVars	EnvironmentVar[]	optional	Environment variables required for worker startup.
expose	Expose	optional	Port information exposed by the worker. If serviceMode=ALB, this is mandatory for your service's HTTP port. If serviceMode=ALB or CUSTOM, this is optional for your service's health check port.

Example — curl

curl --request POST \
--url 'https://api.yottalabs.ai/openapi/v1/elastic/deploy/create' \
--header 'Content-Type: application/json' \
--header 'X-API-Key: <YOUR_API_KEY>' \
--data '{
    "name": "Llama-3.2-3B",
    "image": "vllm/vllm-openai:latest",
    "resources": [
        {
            "region": "us-west-1",
            "gpuType": "NVIDIA_L4_24G",
            "gpuCount": 1
        },
        {
            "region": "us-central-3",
            "gpuType": "NVIDIA_GeForce_RTX_5090_32G",
            "gpuCount": 1
        }
    ],
    "minSingleCardVramInGb": 32,
    "minSingleCardVcpu": 15,
    "minSingleCardRamInGb": 124,
    "containerVolumeInGb": 120,
    "workers": 1,
    "serviceMode": "ALB",
    "initializationCommand": "vllm serve meta-llama/Llama-3.2-3B-Instruct --served-model-name meta-llama/Llama-3.2-3B-Instruct --max-model-len 20480 --port 8000 --dtype half --gpu-memory-utilization 0.90 --tensor-parallel-size 1 --chat-template /vllm-workspace/examples/tool_chat_template_llama3.2_json.jinja",
    "environmentVars": [
        {
            "key": "HUGGING_FACE_HUB_TOKEN",
            "value": "<Your-Hugging_Face_Hub_Token>"
        },
        {
            "key": "VLLM_PORT",
            "value": "8000"
        },
        {
            "key": "CUDA_VISIBLE_DEVICES",
            "value": "0"
        }
    ],
    "expose": {
        "port": 8000,
        "protocol": "http"
    }
}'

Response

Returns an object containing data of type ElasticDeploymentDetail. (See Common Models below)

---

List Elastic Deployments

Endpoint: /openapi/v1/elastic/deploy/list

Method: GET

Description: Lists all Elastic Deployments under the current organization.

Authorizations

Name	Location	Type	Required	Description
X-API-Key	header	string	required	Your API Key

Query Parameters

Name	Type	Required	Description
statusList	string[]	optional	Filter by ElasticDeploymentStatus. Multiple values allowed: statusList=INITIALIZING&statusList=RUNNING&statusList=STOPPED

Example — curl

curl --request GET \
--url 'https://api.yottalabs.ai/openapi/v1/elastic/deploy/list?statusList=INITIALIZING&statusList=RUNNING&statusList=STOPPED' \
--header 'Content-Type: application/json' \
--header 'X-API-Key: <YOUR_API_KEY>'

Response

Returns an object containing data as a list of ElasticDeploymentDetail. (See Common Models below)

---

Get Elastic Deployment Details

Endpoint: /openapi/v1/elastic/deploy/{id}

Method: GET

Description: Retrieves details of a specific Elastic Deployment.

Authorizations

Name	Location	Type	Required	Description
X-API-Key	header	string	required	Your API Key

Path Parameters

Name	Type	Required	Description
id	string	required	Elastic Deployment ID

Example — curl

curl --request GET \
--url 'https://api.yottalabs.ai/openapi/v1/elastic/deploy/384881505694897047' \
--header 'Content-Type: application/json' \
--header 'X-API-Key: <YOUR_API_KEY>'

Response

Returns an object containing data of type ElasticDeploymentDetail. (See Common Models below)

---

Update Elastic Deployment

Endpoint: /openapi/v1/elastic/deploy/{id}/update

Method: POST

Description: Updates a specific Elastic Deployment. Currently, only Elastic Deployments in the STOPPED state can be updated.

Authorizations

Name	Location	Type	Required	Description
X-API-Key	header	string	required	Your API Key

Path Parameters

Name	Type	Required	Description
id	string	required	Elastic Deployment ID

Request Body (JSON)

Name	Type	Required	Description
name	string	required	Elastic Deployment name
resources	Resource[]	required	GPU-related resources. Workers will be deployed on instances satisfying the specified resources. Note: Each Resource must specify the same gpuCount.
minSingleCardVramInGb	number	optional	Minimum VRAM required per single card (GB)
minSingleCardVcpu	number	optional	Minimum VCPU required per single card
minSingleCardRamInGb	number	optional	Minimum RAM required per single card (GB)
containerVolumeInGb	number	required	Container ephemeral volume size (GB)
credentialId	string	optional	Required if using a private image.
workers	number	required	Number of workers to deploy
initializationCommand	string	optional	Initialization command executed upon worker startup
environmentVars	EnvironmentVar[]	optional	Environment variables required for worker startup.
expose	Expose	optional	Port information exposed by the worker.

Example — curl

curl --request POST \
--url 'https://api.yottalabs.ai/openapi/v1/elastic/deploy/386179372792811712/update' \
--header 'Content-Type: application/json' \
--header 'X-API-Key: <YOUR_API_KEY>' \
--data '{
    "name": "Llama-3.2-3B",
    "resources": [
        {
            "region": "us-central-3",
            "gpuType": "NVIDIA_GeForce_RTX_5090_32G",
            "gpuCount": 1
        }
    ],
    "workers": 1,
    "minSingleCardRamInGb": 43,
    "containerVolumeInGb": 20,
    "initializationCommand": "vllm serve...",
    "environmentVars": [
        { "key": "HUGGING_FACE_HUB_TOKEN", "value": "hf_***" }
    ],
    "expose": {
        "port": 8000,
        "protocol": "http"
    }
  }'

Response

Returns an object containing data of type ElasticDeploymentDetail. (See Common Models below)

---

Stop Elastic Deployment

Endpoint: /openapi/v1/elastic/deploy/{id}/stop

Method: POST

Description: Stops the Elastic Deployment with the specified ID.

Authorizations

Name	Location	Type	Required	Description
X-API-Key	header	string	required	Your API Key

Path Parameters

Name	Type	Required	Description
id	string	required	Elastic Deployment ID

Example — curl

curl --request POST \
--url 'https://api.yottalabs.ai/openapi/v1/elastic/deploy/384812561523806429/stop' \
--header 'Content-Type: application/json' \
--header 'X-API-Key: <YOUR_API_KEY>'

Response

Returns data: null on success.

---

Start Elastic Deployment

Endpoint: /openapi/v1/elastic/deploy/{id}/start

Method: POST

Description: Starts the Elastic Deployment with the specified ID.

Authorizations

Name	Location	Type	Required	Description
X-API-Key	header	string	required	Your API Key

Path Parameters

Name	Type	Required	Description
id	string	required	Elastic Deployment ID

Example — curl

curl --request POST \
--url 'https://api.yottalabs.ai/openapi/v1/elastic/deploy/384812561523806429/start' \
--header 'Content-Type: application/json' \
--header 'X-API-Key: <YOUR_API_KEY>'

Response

Returns data: null on success.

---

Delete Elastic Deployment

Endpoint: /openapi/v1/elastic/deploy/{id}

Method: DELETE

Description: Deletes the Elastic Deployment with the specified ID. Note: The deployment must be in a STOPPED state to be deleted.

Authorizations

Name	Location	Type	Required	Description
X-API-Key	header	string	required	Your API Key

Path Parameters

Name	Type	Required	Description
id	string	required	Elastic Deployment ID

Example — curl

curl --request DELETE \
--url 'https://api.yottalabs.ai/openapi/v1/elastic/deploy/384812561523806429' \
--header 'Content-Type: application/json' \
--header 'X-API-Key: <YOUR_API_KEY>'

Response

Returns data: null on success. If deletion is not allowed (e.g., status is not STOPPED), returns error code 24002.

---

Adjust Elastic Deployment Worker Count

Endpoint: /openapi/v1/elastic/deploy/{id}/workers

Method: POST

Description: Adjusts the number of Workers for the specified Elastic Deployment.

Authorizations

Name	Location	Type	Required	Description
X-API-Key	header	string	required	Your API Key

Path Parameters

Name	Type	Required	Description
id	string	required	Elastic Deployment ID

Request Body (JSON)

Name	Type	Required	Description
workers	number	required	Target worker count

Example — curl

curl --request POST \
--url 'https://api.yottalabs.ai/openapi/v1/elastic/deploy/384812561523806429/workers' \
--header 'Content-Type: application/json' \
--header 'X-API-Key: <YOUR_API_KEY>' \
--data '{
    "workers": 2
}'

Response

Returns data: null on success.

---

List Elastic Deployment Workers

Endpoint: /openapi/v1/elastic/deploy/{id}/workers

Method: GET

Description: Retrieves the list of Workers for the specified Elastic Deployment.

Authorizations

Name	Location	Type	Required	Description
X-API-Key	header	string	required	Your API Key

Path Parameters

Name	Type	Required	Description
id	string	required	Elastic Deployment ID

Query Parameters

Name	Type	Required	Description
statusList	string[]	optional	Filter by WorkerStatus. Multiple values allowed: statusList=INITIALIZING&statusList=RUNNING

Example — curl

curl --request GET \
--url 'https://api.yottalabs.ai/openapi/v1/elastic/deploy/384812561523806429/workers?statusList=INITIALIZE&statusList=RUNNING' \
--header 'Content-Type: application/json' \
--header 'X-API-Key: <YOUR_API_KEY>'

Response

Returns an object containing data as a list of ElasticDeploymentWorkerDetail. (See Common Models below)

---

Common Models

Resource

Name	Type	Required	Description
region	string	required	The region where the worker is deployed.
gpuType	string	required	The type of GPU required by the worker.
gpuCount	number	required	The number of GPU cards per worker.

ElasticDeploymentDetail

Name	Type	Description
id	string	Elastic Deployment ID (e.g., "378888638324150969")
name	string	Elastic Deployment Name (e.g., "Llama-3.2-3B")
creator	string	Creator's email
domain	string	ALB Gateway Domain; only exists for Deployments with ServiceMode=ALB.
imageRegistry	string	Image Registry URL; default is Docker Hub.
image	string	Image Name
resources	ResourceResponse[]	Region and GPU resource configuration.
minSingleCardVramInGb	number	Minimum VRAM required per single card (GB)
minSingleCardVcpu	number	Minimum VCPU required per single card
minSingleCardRamInGb	number	Minimum RAM required per single card (GB)
containerVolumeInGb	number	Container ephemeral volume size (GB)
credentialId	string	Credential ID
initializationCommand	string	Service startup command
environmentVars	EnvironmentVar[]	List of environment variables.
expose	Expose	Port exposure configuration.
totalWorkers	number	Total number of Workers
runningWorkers	number	Number of currently running Workers
cost	string	Accumulated cost
perSecondPrice	string	Price per second
perHourPrice	string	Price per hour
serviceMode	string	ALB, QUEUE, or CUSTOM
status	string	INITIALIZING, RUNNING, STOPPING, STOPPED, FAILED

ResourceResponse

Name	Type	Description
region	string	Worker's Region
regionDisplayName	string	Worker's Region Display Name
gpuType	string	GPU Type
gpuDisplayName	string	GPU Type Display Name
gpuCount	number	GPU Card Count
singleCardVramInGb	number	VRAM per single GPU card (GB)
singleCardVcpu	number	VCPU count per single GPU card
singleCardRamInGb	number	RAM per single GPU card (GB)

EnvironmentVar

Name	Type	Required	Description
key	string	required	Name of the environment variable
value	string	required	Value of the environment variable

Expose

Name	Type	Required	Description
port	number	required	Port number, range: 1 ~ 65535
protocol	string	required	Protocol type, supports: HTTP

ElasticDeploymentWorkerDetail

Name	Type	Description
id	string	Worker ID
region	string	Worker's Region
regionDisplayName	string	Worker's Region Display Name
gpuType	string	GPU Type
gpuDisplayName	string	GPU Type Display Name
gpuCount	number	GPU Card Count
singleCardVramInGb	number	VRAM per single GPU card (GB)
singleCardVcpu	number	VCPU count per single GPU card
singleCardRamInGb	number	RAM per single GPU card (GB)
uptime	string	Worker uptime (milliseconds)
cost	string	Cost incurred up to the current time
status	string	INITIALIZE, RUNNING, TERMINATING, TERMINATED, FAILED

Enums

ServiceMode

Name	Description
ALB	Service provided via HTTP requests, includes load balancing capabilities.
QUEUE	Tasks are submitted to the Yotta Queue service for asynchronous processing.
CUSTOM	Neither HTTP requests nor the Queue service are required.

ElasticDeploymentStatus

Name	Description
INITIALIZING	Initializing, starting up.
RUNNING	Running, service is operating normally.
STOPPING	Stopping.
STOPPED	Stopped.
FAILED	Failed.

WorkerStatus

Name	Description
INITIALIZE	Initializing, starting up.
RUNNING	Running, service is operating normally.
TERMINATING	Terminating.
TERMINATED	Terminated.
FAILED	Failed.

Response Codes

Code	Message	Description
10000	success	Request successful
10001	parameter error	Invalid parameter
24001	Elastic does not exist	Elastic Deployment does not exist
24002	Elastic action not allowed	The state of the Elastic Deployment does not allow this operation
24003	You’ve reached the limit of %s deployment	Organization deployment limit reached
24009	Insufficient GPUs available...	Insufficient GPUs available to start the requested number of workers.