Create or Update Deployment

Creates a new deployment or updates an existing one by name/tag, returning an S3 upload URL when a new file version needs to be uploaded.

PUT /api/v0/deployments/

Create or Update Deployment

cURL

curl --request PUT \
  --url https://console.vast.ai/api/v0/deployments/ \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "name": "my-model",
  "tag": "v2",
  "image": "pytorch/pytorch:2.0.0",
  "search_params": "gpu_ram>=16 num_gpus=1",
  "file_hash": "e3b0c44298fc1c149afb",
  "file_size": 1048576,
  "env": "-e MODEL=llama",
  "storage": 100,
  "ttl": 3600,
  "version_label": "release-1.0",
  "docker_login_user": "<string>",
  "docker_login_pass": "<string>",
  "docker_login_repo": "docker.io",
  "cold_workers": 2,
  "max_workers": 10,
  "min_load": 0,
  "min_cold_load": 0,
  "target_util": 0.9,
  "cold_mult": 3,
  "max_queue_time": 60,
  "target_queue_time": 10,
  "inactivity_timeout": 300,
  "overrecruit_ratio": 1.2,
  "autoscaler_instance": "prod"
}
'

import requests

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

payload = {
    "name": "my-model",
    "tag": "v2",
    "image": "pytorch/pytorch:2.0.0",
    "search_params": "gpu_ram>=16 num_gpus=1",
    "file_hash": "e3b0c44298fc1c149afb",
    "file_size": 1048576,
    "env": "-e MODEL=llama",
    "storage": 100,
    "ttl": 3600,
    "version_label": "release-1.0",
    "docker_login_user": "<string>",
    "docker_login_pass": "<string>",
    "docker_login_repo": "docker.io",
    "cold_workers": 2,
    "max_workers": 10,
    "min_load": 0,
    "min_cold_load": 0,
    "target_util": 0.9,
    "cold_mult": 3,
    "max_queue_time": 60,
    "target_queue_time": 10,
    "inactivity_timeout": 300,
    "overrecruit_ratio": 1.2,
    "autoscaler_instance": "prod"
}
headers = {
    "Authorization": "Bearer <token>",
    "Content-Type": "application/json"
}

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

print(response.text)

const options = {
  method: 'PUT',
  headers: {Authorization: 'Bearer <token>', 'Content-Type': 'application/json'},
  body: JSON.stringify({
    name: 'my-model',
    tag: 'v2',
    image: 'pytorch/pytorch:2.0.0',
    search_params: 'gpu_ram>=16 num_gpus=1',
    file_hash: 'e3b0c44298fc1c149afb',
    file_size: 1048576,
    env: '-e MODEL=llama',
    storage: 100,
    ttl: 3600,
    version_label: 'release-1.0',
    docker_login_user: '<string>',
    docker_login_pass: '<string>',
    docker_login_repo: 'docker.io',
    cold_workers: 2,
    max_workers: 10,
    min_load: 0,
    min_cold_load: 0,
    target_util: 0.9,
    cold_mult: 3,
    max_queue_time: 60,
    target_queue_time: 10,
    inactivity_timeout: 300,
    overrecruit_ratio: 1.2,
    autoscaler_instance: 'prod'
  })
};

fetch('https://console.vast.ai/api/v0/deployments/', 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/deployments/",
  CURLOPT_RETURNTRANSFER => true,
  CURLOPT_ENCODING => "",
  CURLOPT_MAXREDIRS => 10,
  CURLOPT_TIMEOUT => 30,
  CURLOPT_HTTP_VERSION => CURL_HTTP_VERSION_1_1,
  CURLOPT_CUSTOMREQUEST => "PUT",
  CURLOPT_POSTFIELDS => json_encode([
    'name' => 'my-model',
    'tag' => 'v2',
    'image' => 'pytorch/pytorch:2.0.0',
    'search_params' => 'gpu_ram>=16 num_gpus=1',
    'file_hash' => 'e3b0c44298fc1c149afb',
    'file_size' => 1048576,
    'env' => '-e MODEL=llama',
    'storage' => 100,
    'ttl' => 3600,
    'version_label' => 'release-1.0',
    'docker_login_user' => '<string>',
    'docker_login_pass' => '<string>',
    'docker_login_repo' => 'docker.io',
    'cold_workers' => 2,
    'max_workers' => 10,
    'min_load' => 0,
    'min_cold_load' => 0,
    'target_util' => 0.9,
    'cold_mult' => 3,
    'max_queue_time' => 60,
    'target_queue_time' => 10,
    'inactivity_timeout' => 300,
    'overrecruit_ratio' => 1.2,
    'autoscaler_instance' => 'prod'
  ]),
  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/deployments/"

	payload := strings.NewReader("{\n  \"name\": \"my-model\",\n  \"tag\": \"v2\",\n  \"image\": \"pytorch/pytorch:2.0.0\",\n  \"search_params\": \"gpu_ram>=16 num_gpus=1\",\n  \"file_hash\": \"e3b0c44298fc1c149afb\",\n  \"file_size\": 1048576,\n  \"env\": \"-e MODEL=llama\",\n  \"storage\": 100,\n  \"ttl\": 3600,\n  \"version_label\": \"release-1.0\",\n  \"docker_login_user\": \"<string>\",\n  \"docker_login_pass\": \"<string>\",\n  \"docker_login_repo\": \"docker.io\",\n  \"cold_workers\": 2,\n  \"max_workers\": 10,\n  \"min_load\": 0,\n  \"min_cold_load\": 0,\n  \"target_util\": 0.9,\n  \"cold_mult\": 3,\n  \"max_queue_time\": 60,\n  \"target_queue_time\": 10,\n  \"inactivity_timeout\": 300,\n  \"overrecruit_ratio\": 1.2,\n  \"autoscaler_instance\": \"prod\"\n}")

	req, _ := http.NewRequest("PUT", 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.put("https://console.vast.ai/api/v0/deployments/")
  .header("Authorization", "Bearer <token>")
  .header("Content-Type", "application/json")
  .body("{\n  \"name\": \"my-model\",\n  \"tag\": \"v2\",\n  \"image\": \"pytorch/pytorch:2.0.0\",\n  \"search_params\": \"gpu_ram>=16 num_gpus=1\",\n  \"file_hash\": \"e3b0c44298fc1c149afb\",\n  \"file_size\": 1048576,\n  \"env\": \"-e MODEL=llama\",\n  \"storage\": 100,\n  \"ttl\": 3600,\n  \"version_label\": \"release-1.0\",\n  \"docker_login_user\": \"<string>\",\n  \"docker_login_pass\": \"<string>\",\n  \"docker_login_repo\": \"docker.io\",\n  \"cold_workers\": 2,\n  \"max_workers\": 10,\n  \"min_load\": 0,\n  \"min_cold_load\": 0,\n  \"target_util\": 0.9,\n  \"cold_mult\": 3,\n  \"max_queue_time\": 60,\n  \"target_queue_time\": 10,\n  \"inactivity_timeout\": 300,\n  \"overrecruit_ratio\": 1.2,\n  \"autoscaler_instance\": \"prod\"\n}")
  .asString();

require 'uri'
require 'net/http'

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

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

request = Net::HTTP::Put.new(url)
request["Authorization"] = 'Bearer <token>'
request["Content-Type"] = 'application/json'
request.body = "{\n  \"name\": \"my-model\",\n  \"tag\": \"v2\",\n  \"image\": \"pytorch/pytorch:2.0.0\",\n  \"search_params\": \"gpu_ram>=16 num_gpus=1\",\n  \"file_hash\": \"e3b0c44298fc1c149afb\",\n  \"file_size\": 1048576,\n  \"env\": \"-e MODEL=llama\",\n  \"storage\": 100,\n  \"ttl\": 3600,\n  \"version_label\": \"release-1.0\",\n  \"docker_login_user\": \"<string>\",\n  \"docker_login_pass\": \"<string>\",\n  \"docker_login_repo\": \"docker.io\",\n  \"cold_workers\": 2,\n  \"max_workers\": 10,\n  \"min_load\": 0,\n  \"min_cold_load\": 0,\n  \"target_util\": 0.9,\n  \"cold_mult\": 3,\n  \"max_queue_time\": 60,\n  \"target_queue_time\": 10,\n  \"inactivity_timeout\": 300,\n  \"overrecruit_ratio\": 1.2,\n  \"autoscaler_instance\": \"prod\"\n}"

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

200

{
  "success": true,
  "action": "created",
  "deployment_id": 5,
  "endpoint_id": 10,
  "upload_url": "https://s3.amazonaws.com/bucket/...",
  "upload_fields": {
    "key": "1/myapp/abc123",
    "AWSAccessKeyId": "..."
  }
}

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

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

Authorizations

Authorization · string · header · required

API key must be provided in the Authorization header

Body

application/json

name · string

Deployment name (letters, numbers, hyphens, dots, underscores; must start with letter or number; max 128 chars).

Example:

"my-model"

tag · string

Deployment tag used to differentiate variants of the same named deployment; defaults to 'default'.

Example:

"v2"

image · string

Docker image to run on worker instances; required.

Example:

"pytorch/pytorch:2.0.0"

search_params · string

GPU offer filter expression for selecting worker machines.

Example:

"gpu_ram>=16 num_gpus=1"

file_hash · string

SHA-256 (or similar) hash of the deployment artifact to upload; required and used as a cache key.

Example:

"e3b0c44298fc1c149afb"

file_size · integer

Size in bytes of the deployment artifact; required when file_hash is provided.

Example:

1048576

env · string

Environment variable string to inject into workers (e.g. '-e FOO=bar').

Example:

"-e MODEL=llama"

storage · number

Disk storage to allocate per worker in GB; must be positive; defaults to 50.

Example:

100

ttl · number

Seconds after the last heartbeat before the deployment is automatically deleted; negative disables auto-expiry.

Example:

3600

version_label · string

Human-readable label for this artifact version; defaults to first 6 chars of file_hash.

Example:

"release-1.0"

docker_login_user · string

Docker registry username for pulling private images.

docker_login_pass · string

Docker registry password for pulling private images.

docker_login_repo · string

Docker registry hostname/repo to authenticate against.

Example:

"docker.io"

cold_workers · integer

Fixed number of cold standby workers for the backing endpoint.

Example:

2

max_workers · integer

Maximum number of workers for the backing endpoint.

Example:

10

min_load · number

Minimum load threshold for the backing endpoint.

Example:

0

min_cold_load · number

Minimum cold load threshold for the backing endpoint.

Example:

0

target_util · number

Target utilization ratio (0-1) for the backing endpoint.

Example:

0.9

cold_mult · number

Cold-standby multiplier for the backing endpoint.

Example:

3

max_queue_time · number

Maximum acceptable queue time in seconds.

Example:

60

target_queue_time · number

Target queue time in seconds.

Example:

10

inactivity_timeout · number

Seconds of inactivity before the endpoint may be suspended.

Example:

300

overrecruit_ratio · number

Ratio by which to over-recruit workers for traffic spike absorption.

Example:

1.2

autoscaler_instance · string

Autoscaler deployment environment.

Example:

"prod"

Response

200

application/json

Returns {success: true, action: 'created'|'soft_update'|'autoscale_update'|'exists', deployment_id, endpoint_id, upload_url?, upload_fields?, evicted_versions?}

success · boolean

Always true on success.

action · string

One of: created, soft_update, autoscale_update, exists.

deployment_id · integer

ID of the deployment.

endpoint_id · integer

ID of the associated endpoint.

upload_url · string

Presigned S3 URL to upload the deployment blob (present when a new version needs to be uploaded).

upload_fields · object

Additional fields required for the S3 multipart POST upload (present with upload_url).

evicted_versions · object[]

List of evicted version objects ({id, file_hash, version_label, file_size}) if LRU eviction occurred.