create workergroup

Creates a new workergroup configuration that manages worker instances for a serverless endpoint.

CLI Usage: vastai create workergroup --template_hash <hash> --endpoint_name <name> [options]

POST /api/v0/workergroups

create workergroup

cURL

curl --request POST \
  --url https://console.vast.ai/api/v0/workergroups \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "endpoint_name": "vLLM-Qwen3-8B",
  "endpoint_id": 123,
  "template_hash": "abc123def456",
  "template_id": 456,
  "search_params": "gpu_name=RTX_3090 rentable=true",
  "launch_args": "--env VAR=value",
  "min_load": 1,
  "target_util": 0.9,
  "cold_mult": 3,
  "cold_workers": 3,
  "max_workers": 20,
  "test_workers": 3,
  "gpu_ram": 24
}
'

import requests

url = "https://console.vast.ai/api/v0/workergroups"

payload = {
    "endpoint_name": "vLLM-Qwen3-8B",
    "endpoint_id": 123,
    "template_hash": "abc123def456",
    "template_id": 456,
    "search_params": "gpu_name=RTX_3090 rentable=true",
    "launch_args": "--env VAR=value",
    "min_load": 1,
    "target_util": 0.9,
    "cold_mult": 3,
    "cold_workers": 3,
    "max_workers": 20,
    "test_workers": 3,
    "gpu_ram": 24
}
headers = {
    "Authorization": "Bearer <token>",
    "Content-Type": "application/json"
}

response = requests.post(url, json=payload, headers=headers)

print(response.text)

const options = {
  method: 'POST',
  headers: {Authorization: 'Bearer <token>', 'Content-Type': 'application/json'},
  body: JSON.stringify({
    endpoint_name: 'vLLM-Qwen3-8B',
    endpoint_id: 123,
    template_hash: 'abc123def456',
    template_id: 456,
    search_params: 'gpu_name=RTX_3090 rentable=true',
    launch_args: '--env VAR=value',
    min_load: 1,
    target_util: 0.9,
    cold_mult: 3,
    cold_workers: 3,
    max_workers: 20,
    test_workers: 3,
    gpu_ram: 24
  })
};

fetch('https://console.vast.ai/api/v0/workergroups', options)
  .then(res => res.json())
  .then(res => console.log(res))
  .catch(err => console.error(err));

<?php

$curl = curl_init();

curl_setopt_array($curl, [
  CURLOPT_URL => "https://console.vast.ai/api/v0/workergroups",
  CURLOPT_RETURNTRANSFER => true,
  CURLOPT_ENCODING => "",
  CURLOPT_MAXREDIRS => 10,
  CURLOPT_TIMEOUT => 30,
  CURLOPT_HTTP_VERSION => CURL_HTTP_VERSION_1_1,
  CURLOPT_CUSTOMREQUEST => "POST",
  CURLOPT_POSTFIELDS => json_encode([
    'endpoint_name' => 'vLLM-Qwen3-8B',
    'endpoint_id' => 123,
    'template_hash' => 'abc123def456',
    'template_id' => 456,
    'search_params' => 'gpu_name=RTX_3090 rentable=true',
    'launch_args' => '--env VAR=value',
    'min_load' => 1,
    'target_util' => 0.9,
    'cold_mult' => 3,
    'cold_workers' => 3,
    'max_workers' => 20,
    'test_workers' => 3,
    'gpu_ram' => 24
  ]),
  CURLOPT_HTTPHEADER => [
    "Authorization: Bearer <token>",
    "Content-Type: application/json"
  ],
]);

$response = curl_exec($curl);
$err = curl_error($curl);

curl_close($curl);

if ($err) {
  echo "cURL Error #:" . $err;
} else {
  echo $response;
}

package main

import (
	"fmt"
	"strings"
	"net/http"
	"io"
)

func main() {

	url := "https://console.vast.ai/api/v0/workergroups"

	payload := strings.NewReader("{\n  \"endpoint_name\": \"vLLM-Qwen3-8B\",\n  \"endpoint_id\": 123,\n  \"template_hash\": \"abc123def456\",\n  \"template_id\": 456,\n  \"search_params\": \"gpu_name=RTX_3090 rentable=true\",\n  \"launch_args\": \"--env VAR=value\",\n  \"min_load\": 1,\n  \"target_util\": 0.9,\n  \"cold_mult\": 3,\n  \"cold_workers\": 3,\n  \"max_workers\": 20,\n  \"test_workers\": 3,\n  \"gpu_ram\": 24\n}")

	req, _ := http.NewRequest("POST", url, payload)

	req.Header.Add("Authorization", "Bearer <token>")
	req.Header.Add("Content-Type", "application/json")

	res, _ := http.DefaultClient.Do(req)

	defer res.Body.Close()
	body, _ := io.ReadAll(res.Body)

	fmt.Println(string(body))

}

HttpResponse<String> response = Unirest.post("https://console.vast.ai/api/v0/workergroups")
  .header("Authorization", "Bearer <token>")
  .header("Content-Type", "application/json")
  .body("{\n  \"endpoint_name\": \"vLLM-Qwen3-8B\",\n  \"endpoint_id\": 123,\n  \"template_hash\": \"abc123def456\",\n  \"template_id\": 456,\n  \"search_params\": \"gpu_name=RTX_3090 rentable=true\",\n  \"launch_args\": \"--env VAR=value\",\n  \"min_load\": 1,\n  \"target_util\": 0.9,\n  \"cold_mult\": 3,\n  \"cold_workers\": 3,\n  \"max_workers\": 20,\n  \"test_workers\": 3,\n  \"gpu_ram\": 24\n}")
  .asString();

require 'uri'
require 'net/http'

url = URI("https://console.vast.ai/api/v0/workergroups")

http = Net::HTTP.new(url.host, url.port)
http.use_ssl = true

request = Net::HTTP::Post.new(url)
request["Authorization"] = 'Bearer <token>'
request["Content-Type"] = 'application/json'
request.body = "{\n  \"endpoint_name\": \"vLLM-Qwen3-8B\",\n  \"endpoint_id\": 123,\n  \"template_hash\": \"abc123def456\",\n  \"template_id\": 456,\n  \"search_params\": \"gpu_name=RTX_3090 rentable=true\",\n  \"launch_args\": \"--env VAR=value\",\n  \"min_load\": 1,\n  \"target_util\": 0.9,\n  \"cold_mult\": 3,\n  \"cold_workers\": 3,\n  \"max_workers\": 20,\n  \"test_workers\": 3,\n  \"gpu_ram\": 24\n}"

response = http.request(request)
puts response.read_body

200

{
  "success": true,
  "id": 789
}

{
  "success": false,
  "error": "invalid_args",
  "msg": "Please assign your workergroup to a valid endpoint identifier"
}

{
  "success": false,
  "error": "<string>",
  "msg": "<string>"
}

{
  "detail": "API requests too frequent endpoint threshold=4.0"
}

Authorizations

Authorization · string · header · required

API key must be provided in the Authorization header

Body

application/json

endpoint_name · string

Name of the endpoint group

Example:

"vLLM-Qwen3-8B"

endpoint_id · integer

ID of existing endpoint group (alternative to endpoint_name)

Example:

123

template_hash · string

Hash ID of template to use for worker instances

Example:

"abc123def456"

template_id · integer

ID of template (alternative to template_hash)

Example:

456

search_params · string · default: verified=true rentable=true rented=false

Search query for finding worker instances (alternative to template)

Example:

"gpu_name=RTX_3090 rentable=true"

launch_args · string

Additional launch arguments for worker instances

Example:

"--env VAR=value"

min_load · number · default: 1

Minimum load threshold for scaling

Example:

1

target_util · number · default: 0.9

Target GPU utilization

Example:

0.9

cold_mult · number · default: 3

Cold start multiplier

Example:

3

cold_workers · integer · default: 3

Number of cold workers to maintain

Example:

3

max_workers · integer · default: 20

Maximum number of worker instances

Example:

20

test_workers · integer · default: 3

Number of test workers

Example:

3

gpu_ram · integer · default: 24

Minimum GPU RAM in GB

Example:

24

Response

200

application/json

Successfully created workergroup

success · boolean

Example:

true

id · integer

ID of created autoscaling job

Example:

789