Deploy Create Llm

curl --request POST \
  --url https://api.deepinfra.com/deploy/llm \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "model_name": "<string>",
  "num_gpus": 1,
  "max_batch_size": 96,
  "hf": {
    "repo": "<string>",
    "revision": "<string>",
    "token": "<string>"
  },
  "base_model": "<string>",
  "container_image": "<string>",
  "settings": {
    "min_instances": 1,
    "max_instances": 1
  },
  "extra_args": [
    "<string>"
  ],
  "standard_args": {
    "max_context_size": 5000000,
    "max_concurrent_requests": 512,
    "gpu_memory_fraction": 0.735,
    "max_prefill_tokens": 65792,
    "enable_prefix_caching": true
  },
  "preset_id": "<string>"
}
'

import requests

url = "https://api.deepinfra.com/deploy/llm"

payload = {
    "model_name": "<string>",
    "num_gpus": 1,
    "max_batch_size": 96,
    "hf": {
        "repo": "<string>",
        "revision": "<string>",
        "token": "<string>"
    },
    "base_model": "<string>",
    "container_image": "<string>",
    "settings": {
        "min_instances": 1,
        "max_instances": 1
    },
    "extra_args": ["<string>"],
    "standard_args": {
        "max_context_size": 5000000,
        "max_concurrent_requests": 512,
        "gpu_memory_fraction": 0.735,
        "max_prefill_tokens": 65792,
        "enable_prefix_caching": True
    },
    "preset_id": "<string>"
}
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({
    model_name: '<string>',
    num_gpus: 1,
    max_batch_size: 96,
    hf: {repo: '<string>', revision: '<string>', token: '<string>'},
    base_model: '<string>',
    container_image: '<string>',
    settings: {min_instances: 1, max_instances: 1},
    extra_args: ['<string>'],
    standard_args: {
      max_context_size: 5000000,
      max_concurrent_requests: 512,
      gpu_memory_fraction: 0.735,
      max_prefill_tokens: 65792,
      enable_prefix_caching: true
    },
    preset_id: '<string>'
  })
};

fetch('https://api.deepinfra.com/deploy/llm', 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://api.deepinfra.com/deploy/llm",
  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([
    'model_name' => '<string>',
    'num_gpus' => 1,
    'max_batch_size' => 96,
    'hf' => [
        'repo' => '<string>',
        'revision' => '<string>',
        'token' => '<string>'
    ],
    'base_model' => '<string>',
    'container_image' => '<string>',
    'settings' => [
        'min_instances' => 1,
        'max_instances' => 1
    ],
    'extra_args' => [
        '<string>'
    ],
    'standard_args' => [
        'max_context_size' => 5000000,
        'max_concurrent_requests' => 512,
        'gpu_memory_fraction' => 0.735,
        'max_prefill_tokens' => 65792,
        'enable_prefix_caching' => true
    ],
    'preset_id' => '<string>'
  ]),
  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://api.deepinfra.com/deploy/llm"

	payload := strings.NewReader("{\n  \"model_name\": \"<string>\",\n  \"num_gpus\": 1,\n  \"max_batch_size\": 96,\n  \"hf\": {\n    \"repo\": \"<string>\",\n    \"revision\": \"<string>\",\n    \"token\": \"<string>\"\n  },\n  \"base_model\": \"<string>\",\n  \"container_image\": \"<string>\",\n  \"settings\": {\n    \"min_instances\": 1,\n    \"max_instances\": 1\n  },\n  \"extra_args\": [\n    \"<string>\"\n  ],\n  \"standard_args\": {\n    \"max_context_size\": 5000000,\n    \"max_concurrent_requests\": 512,\n    \"gpu_memory_fraction\": 0.735,\n    \"max_prefill_tokens\": 65792,\n    \"enable_prefix_caching\": true\n  },\n  \"preset_id\": \"<string>\"\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://api.deepinfra.com/deploy/llm")
  .header("Authorization", "Bearer <token>")
  .header("Content-Type", "application/json")
  .body("{\n  \"model_name\": \"<string>\",\n  \"num_gpus\": 1,\n  \"max_batch_size\": 96,\n  \"hf\": {\n    \"repo\": \"<string>\",\n    \"revision\": \"<string>\",\n    \"token\": \"<string>\"\n  },\n  \"base_model\": \"<string>\",\n  \"container_image\": \"<string>\",\n  \"settings\": {\n    \"min_instances\": 1,\n    \"max_instances\": 1\n  },\n  \"extra_args\": [\n    \"<string>\"\n  ],\n  \"standard_args\": {\n    \"max_context_size\": 5000000,\n    \"max_concurrent_requests\": 512,\n    \"gpu_memory_fraction\": 0.735,\n    \"max_prefill_tokens\": 65792,\n    \"enable_prefix_caching\": true\n  },\n  \"preset_id\": \"<string>\"\n}")
  .asString();

require 'uri'
require 'net/http'

url = URI("https://api.deepinfra.com/deploy/llm")

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  \"model_name\": \"<string>\",\n  \"num_gpus\": 1,\n  \"max_batch_size\": 96,\n  \"hf\": {\n    \"repo\": \"<string>\",\n    \"revision\": \"<string>\",\n    \"token\": \"<string>\"\n  },\n  \"base_model\": \"<string>\",\n  \"container_image\": \"<string>\",\n  \"settings\": {\n    \"min_instances\": 1,\n    \"max_instances\": 1\n  },\n  \"extra_args\": [\n    \"<string>\"\n  ],\n  \"standard_args\": {\n    \"max_context_size\": 5000000,\n    \"max_concurrent_requests\": 512,\n    \"gpu_memory_fraction\": 0.735,\n    \"max_prefill_tokens\": 65792,\n    \"enable_prefix_caching\": true\n  },\n  \"preset_id\": \"<string>\"\n}"

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

{
  "deploy_id": "<string>",
  "model_name": "<string>",
  "version": "<string>",
  "task": "<string>",
  "status": "<string>",
  "fail_reason": "<string>",
  "created_at": "<string>",
  "updated_at": "<string>",
  "type": "legacy",
  "instances": {
    "running": 123,
    "pending": 123
  },
  "config": {
    "num_gpus": 123,
    "max_batch_size": 123,
    "weights": {
      "repo": "<string>",
      "revision": "<string>",
      "token": "<string>"
    }
  },
  "settings": {
    "min_instances": 1,
    "max_instances": 1
  },
  "standard_args": {
    "max_context_size": 5000000,
    "max_concurrent_requests": 512,
    "gpu_memory_fraction": 0.735,
    "max_prefill_tokens": 65792,
    "enable_prefix_caching": true
  },
  "extra_args": [
    "<string>"
  ],
  "update_error": "<string>",
  "rollout": {
    "up_to_date": 123,
    "outdated": 123,
    "booting": 123
  }
}

{
  "error": "<string>"
}

{
  "error": "<string>"
}

{
  "error": "<string>"
}

{
  "error": "<string>"
}

POST

deploy

llm

Deploy Create Llm

curl --request POST \
  --url https://api.deepinfra.com/deploy/llm \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "model_name": "<string>",
  "num_gpus": 1,
  "max_batch_size": 96,
  "hf": {
    "repo": "<string>",
    "revision": "<string>",
    "token": "<string>"
  },
  "base_model": "<string>",
  "container_image": "<string>",
  "settings": {
    "min_instances": 1,
    "max_instances": 1
  },
  "extra_args": [
    "<string>"
  ],
  "standard_args": {
    "max_context_size": 5000000,
    "max_concurrent_requests": 512,
    "gpu_memory_fraction": 0.735,
    "max_prefill_tokens": 65792,
    "enable_prefix_caching": true
  },
  "preset_id": "<string>"
}
'

import requests

url = "https://api.deepinfra.com/deploy/llm"

payload = {
    "model_name": "<string>",
    "num_gpus": 1,
    "max_batch_size": 96,
    "hf": {
        "repo": "<string>",
        "revision": "<string>",
        "token": "<string>"
    },
    "base_model": "<string>",
    "container_image": "<string>",
    "settings": {
        "min_instances": 1,
        "max_instances": 1
    },
    "extra_args": ["<string>"],
    "standard_args": {
        "max_context_size": 5000000,
        "max_concurrent_requests": 512,
        "gpu_memory_fraction": 0.735,
        "max_prefill_tokens": 65792,
        "enable_prefix_caching": True
    },
    "preset_id": "<string>"
}
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({
    model_name: '<string>',
    num_gpus: 1,
    max_batch_size: 96,
    hf: {repo: '<string>', revision: '<string>', token: '<string>'},
    base_model: '<string>',
    container_image: '<string>',
    settings: {min_instances: 1, max_instances: 1},
    extra_args: ['<string>'],
    standard_args: {
      max_context_size: 5000000,
      max_concurrent_requests: 512,
      gpu_memory_fraction: 0.735,
      max_prefill_tokens: 65792,
      enable_prefix_caching: true
    },
    preset_id: '<string>'
  })
};

fetch('https://api.deepinfra.com/deploy/llm', 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://api.deepinfra.com/deploy/llm",
  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([
    'model_name' => '<string>',
    'num_gpus' => 1,
    'max_batch_size' => 96,
    'hf' => [
        'repo' => '<string>',
        'revision' => '<string>',
        'token' => '<string>'
    ],
    'base_model' => '<string>',
    'container_image' => '<string>',
    'settings' => [
        'min_instances' => 1,
        'max_instances' => 1
    ],
    'extra_args' => [
        '<string>'
    ],
    'standard_args' => [
        'max_context_size' => 5000000,
        'max_concurrent_requests' => 512,
        'gpu_memory_fraction' => 0.735,
        'max_prefill_tokens' => 65792,
        'enable_prefix_caching' => true
    ],
    'preset_id' => '<string>'
  ]),
  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://api.deepinfra.com/deploy/llm"

	payload := strings.NewReader("{\n  \"model_name\": \"<string>\",\n  \"num_gpus\": 1,\n  \"max_batch_size\": 96,\n  \"hf\": {\n    \"repo\": \"<string>\",\n    \"revision\": \"<string>\",\n    \"token\": \"<string>\"\n  },\n  \"base_model\": \"<string>\",\n  \"container_image\": \"<string>\",\n  \"settings\": {\n    \"min_instances\": 1,\n    \"max_instances\": 1\n  },\n  \"extra_args\": [\n    \"<string>\"\n  ],\n  \"standard_args\": {\n    \"max_context_size\": 5000000,\n    \"max_concurrent_requests\": 512,\n    \"gpu_memory_fraction\": 0.735,\n    \"max_prefill_tokens\": 65792,\n    \"enable_prefix_caching\": true\n  },\n  \"preset_id\": \"<string>\"\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://api.deepinfra.com/deploy/llm")
  .header("Authorization", "Bearer <token>")
  .header("Content-Type", "application/json")
  .body("{\n  \"model_name\": \"<string>\",\n  \"num_gpus\": 1,\n  \"max_batch_size\": 96,\n  \"hf\": {\n    \"repo\": \"<string>\",\n    \"revision\": \"<string>\",\n    \"token\": \"<string>\"\n  },\n  \"base_model\": \"<string>\",\n  \"container_image\": \"<string>\",\n  \"settings\": {\n    \"min_instances\": 1,\n    \"max_instances\": 1\n  },\n  \"extra_args\": [\n    \"<string>\"\n  ],\n  \"standard_args\": {\n    \"max_context_size\": 5000000,\n    \"max_concurrent_requests\": 512,\n    \"gpu_memory_fraction\": 0.735,\n    \"max_prefill_tokens\": 65792,\n    \"enable_prefix_caching\": true\n  },\n  \"preset_id\": \"<string>\"\n}")
  .asString();

require 'uri'
require 'net/http'

url = URI("https://api.deepinfra.com/deploy/llm")

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  \"model_name\": \"<string>\",\n  \"num_gpus\": 1,\n  \"max_batch_size\": 96,\n  \"hf\": {\n    \"repo\": \"<string>\",\n    \"revision\": \"<string>\",\n    \"token\": \"<string>\"\n  },\n  \"base_model\": \"<string>\",\n  \"container_image\": \"<string>\",\n  \"settings\": {\n    \"min_instances\": 1,\n    \"max_instances\": 1\n  },\n  \"extra_args\": [\n    \"<string>\"\n  ],\n  \"standard_args\": {\n    \"max_context_size\": 5000000,\n    \"max_concurrent_requests\": 512,\n    \"gpu_memory_fraction\": 0.735,\n    \"max_prefill_tokens\": 65792,\n    \"enable_prefix_caching\": true\n  },\n  \"preset_id\": \"<string>\"\n}"

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

{
  "deploy_id": "<string>",
  "model_name": "<string>",
  "version": "<string>",
  "task": "<string>",
  "status": "<string>",
  "fail_reason": "<string>",
  "created_at": "<string>",
  "updated_at": "<string>",
  "type": "legacy",
  "instances": {
    "running": 123,
    "pending": 123
  },
  "config": {
    "num_gpus": 123,
    "max_batch_size": 123,
    "weights": {
      "repo": "<string>",
      "revision": "<string>",
      "token": "<string>"
    }
  },
  "settings": {
    "min_instances": 1,
    "max_instances": 1
  },
  "standard_args": {
    "max_context_size": 5000000,
    "max_concurrent_requests": 512,
    "gpu_memory_fraction": 0.735,
    "max_prefill_tokens": 65792,
    "enable_prefix_caching": true
  },
  "extra_args": [
    "<string>"
  ],
  "update_error": "<string>",
  "rollout": {
    "up_to_date": 123,
    "outdated": 123,
    "booting": 123
  }
}

{
  "error": "<string>"
}

{
  "error": "<string>"
}

{
  "error": "<string>"
}

{
  "error": "<string>"
}

Authorizations

Authorization

string

header

required

Bearer authentication header of the form Bearer <token>, where <token> is your auth token.

Headers

xi-api-key

string | null

x-api-key

string | null

Body

application/json

model_name

string

required

model name for deepinfra (username/mode-name format)

gpu

enum<string>

required

The type of GPU the deployment is running on.

Available options:

L4-24GB,

L40S-48GB,

A100-80GB,

H100-80GB,

H200-141GB,

B200-180GB,

B300-270GB,

RTXPRO6000-96GB,

other

num_gpus

integer

default:1

Number of GPUs used by one instance

Required range: 1 <= x <= 8

max_batch_size

integer

default:96

Maximum number of concurrent requests

Required range: 1 <= x <= 256

HFWeights · object | null

Show child attributes

base_model

string | null

Base public model

container_image

string | null

Docker image for the deployment (e.g. vllm/vllm-openai:v0.8.4)

settings

ScaleSettings · object | null

Show child attributes

extra_args

string[] | null

Extra command line arguments for custom deployments

standard_args

StandardArgs · object | null

Engine tuning knobs. Values are validated on submission; unsupported or out-of-range values are rejected.

Show child attributes

preset_id

string | null

Apply a stored preset or vLLM recipe by id. A non-empty standard_args or extra_args in this request replaces the preset's whole matching field.

Response

Successful Response

deploy_id

string

required

Deploy Id

Example:

"fkj843kjh8"

model_name

string

required

Model Id from huggingface

Example:

"google/vit-base-patch16-224"

version

string

required

Model version

Example:

"d8b79b422843bd59d628bf25b01aded94a9ec1a9b917e69fe460df9ff39ec42b"

task

string

required

Task

Example:

"image-classification"

status

string

required

Status

Example:

"deployed"

fail_reason

string

required

Failure reason

Example:

"Initialization failed"

created_at

string

required

Created at

Example:

"2021-08-27T17:19:21+00:00"

updated_at

string

required

Updated at

Example:

"2021-08-27T17:19:21+00:00"

type

enum<string>

default:legacy

Available options:

legacy,

llm,

lora,

tts

instances

DeployInstances · object | null

Details about number of instances running right now

Show child attributes

config

DeployLLMConfig · object | null

Immutable deploy configuration

Show child attributes

settings

ScaleSettings · object | null

Scale Settings

Show child attributes

standard_args

StandardArgs · object

Current engine tuning knobs

Show child attributes

extra_args

string[] | null

Current extra engine-specific command-line args (custom-weight deploys only)

update_error

string | null

If the last config update was auto-reverted after an engine crash-loop, the error that caused it (status stays 'running')

rollout

DeployRollout · object | null

Per-instance rollout progress of the current config

Show child attributes

Deploy Llm Presets Deploy List

Logs & Metrics

Utilities

Account

Dedicated Models

Models

LoRA Adapters

Inference

Tokenizer

Chat Completions

Image Generation

Audio

Videos

Text to Speech

GPU Rentals

Text Completions

Embeddings

Authentication

Files & Batches

Agents

Sandboxes

Billing

Deploy Create Llm

Authorizations

Headers

Body

Response