Update AI Config variation

curl --request PATCH \
  --url https://app.launchdarkly.com/api/v2/projects/{projectKey}/ai-configs/{configKey}/variations/{variationKey} \
  --header 'Authorization: <api-key>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "judgeConfiguration": {
    "judges": [
      {
        "samplingRate": 0.36160767,
        "judgeConfigKey": "judgeConfigKey"
      },
      {
        "samplingRate": 0.36160767,
        "judgeConfigKey": "judgeConfigKey"
      }
    ]
  },
  "instructions": "instructions",
  "description": "description",
  "published": true,
  "modelConfigKey": "OpenAI.gpt-4.1",
  "tools": [
    {
      "version": 6,
      "key": "key"
    },
    {
      "version": 6,
      "key": "key"
    }
  ],
  "skills": [
    {
      "version": 1,
      "key": "key"
    },
    {
      "version": 1,
      "key": "key"
    }
  ],
  "toolKeys": [
    "toolKeys",
    "toolKeys"
  ],
  "classifiers": [
    {
      "instructions": "instructions",
      "criteria": "",
      "type": "bool",
      "key": "key"
    },
    {
      "instructions": "instructions",
      "criteria": "",
      "type": "bool",
      "key": "key"
    }
  ],
  "name": "name",
  "messages": [
    {
      "role": "role",
      "content": "content"
    },
    {
      "role": "role",
      "content": "content"
    }
  ],
  "comment": "comment",
  "model": "{}",
  "state": "archived",
  "outputFormat": {
    "key": ""
  },
  "modelConfigVersion": 1
}
'

import requests

url = "https://app.launchdarkly.com/api/v2/projects/{projectKey}/ai-configs/{configKey}/variations/{variationKey}"

payload = {
    "judgeConfiguration": { "judges": [
            {
                "samplingRate": 0.36160767,
                "judgeConfigKey": "judgeConfigKey"
            },
            {
                "samplingRate": 0.36160767,
                "judgeConfigKey": "judgeConfigKey"
            }
        ] },
    "instructions": "instructions",
    "description": "description",
    "published": True,
    "modelConfigKey": "OpenAI.gpt-4.1",
    "tools": [
        {
            "version": 6,
            "key": "key"
        },
        {
            "version": 6,
            "key": "key"
        }
    ],
    "skills": [
        {
            "version": 1,
            "key": "key"
        },
        {
            "version": 1,
            "key": "key"
        }
    ],
    "toolKeys": ["toolKeys", "toolKeys"],
    "classifiers": [
        {
            "instructions": "instructions",
            "criteria": "",
            "type": "bool",
            "key": "key"
        },
        {
            "instructions": "instructions",
            "criteria": "",
            "type": "bool",
            "key": "key"
        }
    ],
    "name": "name",
    "messages": [
        {
            "role": "role",
            "content": "content"
        },
        {
            "role": "role",
            "content": "content"
        }
    ],
    "comment": "comment",
    "model": "{}",
    "state": "archived",
    "outputFormat": { "key": "" },
    "modelConfigVersion": 1
}
headers = {
    "Authorization": "<api-key>",
    "Content-Type": "application/json"
}

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

print(response.text)

const options = {
  method: 'PATCH',
  headers: {Authorization: '<api-key>', 'Content-Type': 'application/json'},
  body: JSON.stringify({
    judgeConfiguration: {
      judges: [
        {samplingRate: 0.36160767, judgeConfigKey: 'judgeConfigKey'},
        {samplingRate: 0.36160767, judgeConfigKey: 'judgeConfigKey'}
      ]
    },
    instructions: 'instructions',
    description: 'description',
    published: true,
    modelConfigKey: 'OpenAI.gpt-4.1',
    tools: [{version: 6, key: 'key'}, {version: 6, key: 'key'}],
    skills: [{version: 1, key: 'key'}, {version: 1, key: 'key'}],
    toolKeys: ['toolKeys', 'toolKeys'],
    classifiers: [
      {instructions: 'instructions', criteria: '', type: 'bool', key: 'key'},
      {instructions: 'instructions', criteria: '', type: 'bool', key: 'key'}
    ],
    name: 'name',
    messages: [{role: 'role', content: 'content'}, {role: 'role', content: 'content'}],
    comment: 'comment',
    model: '{}',
    state: 'archived',
    outputFormat: {key: ''},
    modelConfigVersion: 1
  })
};

fetch('https://app.launchdarkly.com/api/v2/projects/{projectKey}/ai-configs/{configKey}/variations/{variationKey}', 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://app.launchdarkly.com/api/v2/projects/{projectKey}/ai-configs/{configKey}/variations/{variationKey}",
  CURLOPT_RETURNTRANSFER => true,
  CURLOPT_ENCODING => "",
  CURLOPT_MAXREDIRS => 10,
  CURLOPT_TIMEOUT => 30,
  CURLOPT_HTTP_VERSION => CURL_HTTP_VERSION_1_1,
  CURLOPT_CUSTOMREQUEST => "PATCH",
  CURLOPT_POSTFIELDS => json_encode([
    'judgeConfiguration' => [
        'judges' => [
                [
                                'samplingRate' => 0.36160767,
                                'judgeConfigKey' => 'judgeConfigKey'
                ],
                [
                                'samplingRate' => 0.36160767,
                                'judgeConfigKey' => 'judgeConfigKey'
                ]
        ]
    ],
    'instructions' => 'instructions',
    'description' => 'description',
    'published' => true,
    'modelConfigKey' => 'OpenAI.gpt-4.1',
    'tools' => [
        [
                'version' => 6,
                'key' => 'key'
        ],
        [
                'version' => 6,
                'key' => 'key'
        ]
    ],
    'skills' => [
        [
                'version' => 1,
                'key' => 'key'
        ],
        [
                'version' => 1,
                'key' => 'key'
        ]
    ],
    'toolKeys' => [
        'toolKeys',
        'toolKeys'
    ],
    'classifiers' => [
        [
                'instructions' => 'instructions',
                'criteria' => '',
                'type' => 'bool',
                'key' => 'key'
        ],
        [
                'instructions' => 'instructions',
                'criteria' => '',
                'type' => 'bool',
                'key' => 'key'
        ]
    ],
    'name' => 'name',
    'messages' => [
        [
                'role' => 'role',
                'content' => 'content'
        ],
        [
                'role' => 'role',
                'content' => 'content'
        ]
    ],
    'comment' => 'comment',
    'model' => '{}',
    'state' => 'archived',
    'outputFormat' => [
        'key' => ''
    ],
    'modelConfigVersion' => 1
  ]),
  CURLOPT_HTTPHEADER => [
    "Authorization: <api-key>",
    "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://app.launchdarkly.com/api/v2/projects/{projectKey}/ai-configs/{configKey}/variations/{variationKey}"

	payload := strings.NewReader("{\n  \"judgeConfiguration\": {\n    \"judges\": [\n      {\n        \"samplingRate\": 0.36160767,\n        \"judgeConfigKey\": \"judgeConfigKey\"\n      },\n      {\n        \"samplingRate\": 0.36160767,\n        \"judgeConfigKey\": \"judgeConfigKey\"\n      }\n    ]\n  },\n  \"instructions\": \"instructions\",\n  \"description\": \"description\",\n  \"published\": true,\n  \"modelConfigKey\": \"OpenAI.gpt-4.1\",\n  \"tools\": [\n    {\n      \"version\": 6,\n      \"key\": \"key\"\n    },\n    {\n      \"version\": 6,\n      \"key\": \"key\"\n    }\n  ],\n  \"skills\": [\n    {\n      \"version\": 1,\n      \"key\": \"key\"\n    },\n    {\n      \"version\": 1,\n      \"key\": \"key\"\n    }\n  ],\n  \"toolKeys\": [\n    \"toolKeys\",\n    \"toolKeys\"\n  ],\n  \"classifiers\": [\n    {\n      \"instructions\": \"instructions\",\n      \"criteria\": \"\",\n      \"type\": \"bool\",\n      \"key\": \"key\"\n    },\n    {\n      \"instructions\": \"instructions\",\n      \"criteria\": \"\",\n      \"type\": \"bool\",\n      \"key\": \"key\"\n    }\n  ],\n  \"name\": \"name\",\n  \"messages\": [\n    {\n      \"role\": \"role\",\n      \"content\": \"content\"\n    },\n    {\n      \"role\": \"role\",\n      \"content\": \"content\"\n    }\n  ],\n  \"comment\": \"comment\",\n  \"model\": \"{}\",\n  \"state\": \"archived\",\n  \"outputFormat\": {\n    \"key\": \"\"\n  },\n  \"modelConfigVersion\": 1\n}")

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

	req.Header.Add("Authorization", "<api-key>")
	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.patch("https://app.launchdarkly.com/api/v2/projects/{projectKey}/ai-configs/{configKey}/variations/{variationKey}")
  .header("Authorization", "<api-key>")
  .header("Content-Type", "application/json")
  .body("{\n  \"judgeConfiguration\": {\n    \"judges\": [\n      {\n        \"samplingRate\": 0.36160767,\n        \"judgeConfigKey\": \"judgeConfigKey\"\n      },\n      {\n        \"samplingRate\": 0.36160767,\n        \"judgeConfigKey\": \"judgeConfigKey\"\n      }\n    ]\n  },\n  \"instructions\": \"instructions\",\n  \"description\": \"description\",\n  \"published\": true,\n  \"modelConfigKey\": \"OpenAI.gpt-4.1\",\n  \"tools\": [\n    {\n      \"version\": 6,\n      \"key\": \"key\"\n    },\n    {\n      \"version\": 6,\n      \"key\": \"key\"\n    }\n  ],\n  \"skills\": [\n    {\n      \"version\": 1,\n      \"key\": \"key\"\n    },\n    {\n      \"version\": 1,\n      \"key\": \"key\"\n    }\n  ],\n  \"toolKeys\": [\n    \"toolKeys\",\n    \"toolKeys\"\n  ],\n  \"classifiers\": [\n    {\n      \"instructions\": \"instructions\",\n      \"criteria\": \"\",\n      \"type\": \"bool\",\n      \"key\": \"key\"\n    },\n    {\n      \"instructions\": \"instructions\",\n      \"criteria\": \"\",\n      \"type\": \"bool\",\n      \"key\": \"key\"\n    }\n  ],\n  \"name\": \"name\",\n  \"messages\": [\n    {\n      \"role\": \"role\",\n      \"content\": \"content\"\n    },\n    {\n      \"role\": \"role\",\n      \"content\": \"content\"\n    }\n  ],\n  \"comment\": \"comment\",\n  \"model\": \"{}\",\n  \"state\": \"archived\",\n  \"outputFormat\": {\n    \"key\": \"\"\n  },\n  \"modelConfigVersion\": 1\n}")
  .asString();

require 'uri'
require 'net/http'

url = URI("https://app.launchdarkly.com/api/v2/projects/{projectKey}/ai-configs/{configKey}/variations/{variationKey}")

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

request = Net::HTTP::Patch.new(url)
request["Authorization"] = '<api-key>'
request["Content-Type"] = 'application/json'
request.body = "{\n  \"judgeConfiguration\": {\n    \"judges\": [\n      {\n        \"samplingRate\": 0.36160767,\n        \"judgeConfigKey\": \"judgeConfigKey\"\n      },\n      {\n        \"samplingRate\": 0.36160767,\n        \"judgeConfigKey\": \"judgeConfigKey\"\n      }\n    ]\n  },\n  \"instructions\": \"instructions\",\n  \"description\": \"description\",\n  \"published\": true,\n  \"modelConfigKey\": \"OpenAI.gpt-4.1\",\n  \"tools\": [\n    {\n      \"version\": 6,\n      \"key\": \"key\"\n    },\n    {\n      \"version\": 6,\n      \"key\": \"key\"\n    }\n  ],\n  \"skills\": [\n    {\n      \"version\": 1,\n      \"key\": \"key\"\n    },\n    {\n      \"version\": 1,\n      \"key\": \"key\"\n    }\n  ],\n  \"toolKeys\": [\n    \"toolKeys\",\n    \"toolKeys\"\n  ],\n  \"classifiers\": [\n    {\n      \"instructions\": \"instructions\",\n      \"criteria\": \"\",\n      \"type\": \"bool\",\n      \"key\": \"key\"\n    },\n    {\n      \"instructions\": \"instructions\",\n      \"criteria\": \"\",\n      \"type\": \"bool\",\n      \"key\": \"key\"\n    }\n  ],\n  \"name\": \"name\",\n  \"messages\": [\n    {\n      \"role\": \"role\",\n      \"content\": \"content\"\n    },\n    {\n      \"role\": \"role\",\n      \"content\": \"content\"\n    }\n  ],\n  \"comment\": \"comment\",\n  \"model\": \"{}\",\n  \"state\": \"archived\",\n  \"outputFormat\": {\n    \"key\": \"\"\n  },\n  \"modelConfigVersion\": 1\n}"

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

{
  "judgeConfiguration": {
    "judges": [
      {
        "samplingRate": 0.36160767,
        "judgeConfigKey": "judgeConfigKey"
      },
      {
        "samplingRate": 0.36160767,
        "judgeConfigKey": "judgeConfigKey"
      }
    ]
  },
  "instructions": "instructions",
  "color": "color",
  "_links": {
    "parent": {
      "href": "href",
      "type": "type"
    }
  },
  "description": "description",
  "modelConfigKey": "OpenAI.gpt-4.1",
  "judgingConfigKeys": [
    "judgingConfigKeys",
    "judgingConfigKeys"
  ],
  "version": 5,
  "tools": [
    {
      "customParameters": {
        "key": ""
      },
      "version": 7,
      "key": "key"
    },
    {
      "customParameters": {
        "key": ""
      },
      "version": 7,
      "key": "key"
    }
  ],
  "skills": [
    {
      "version": 9,
      "key": "key"
    },
    {
      "version": 9,
      "key": "key"
    }
  ],
  "createdAt": 1,
  "_publishedAt": 2,
  "_archivedAt": 5,
  "classifiers": [
    {
      "instructions": "instructions",
      "criteria": "",
      "type": "bool",
      "key": "key"
    },
    {
      "instructions": "instructions",
      "criteria": "",
      "type": "bool",
      "key": "key"
    }
  ],
  "name": "name",
  "messages": [
    {
      "role": "role",
      "content": "content"
    },
    {
      "role": "role",
      "content": "content"
    }
  ],
  "comment": "comment",
  "model": "{}",
  "_id": "_id",
  "state": "published",
  "outputFormat": {
    "key": ""
  },
  "key": "key",
  "modelConfigVersion": 6
}

{
  "code": "code",
  "message": "message"
}

{
  "code": "code",
  "message": "message"
}

{
  "code": "code",
  "message": "message"
}

{
  "code": "code",
  "message": "message"
}

Update AI Config variation

Edit an existing variation of an AI Config. This creates a new version of the variation.

The request body must be a JSON object of the fields to update. The values you include replace the existing values for the fields.

If the variation is in use in a project environment where approval settings require review for this AI Config, this endpoint returns 403 and the change must be submitted through an approval request.

Here’s an example:

  {
    "messages": [
      {
        "role": "system",
        "content": "The new message"
      }
    ]
  }

PATCH

api

projects

{projectKey}

ai-configs

{configKey}

variations

{variationKey}

Update AI Config variation

curl --request PATCH \
  --url https://app.launchdarkly.com/api/v2/projects/{projectKey}/ai-configs/{configKey}/variations/{variationKey} \
  --header 'Authorization: <api-key>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "judgeConfiguration": {
    "judges": [
      {
        "samplingRate": 0.36160767,
        "judgeConfigKey": "judgeConfigKey"
      },
      {
        "samplingRate": 0.36160767,
        "judgeConfigKey": "judgeConfigKey"
      }
    ]
  },
  "instructions": "instructions",
  "description": "description",
  "published": true,
  "modelConfigKey": "OpenAI.gpt-4.1",
  "tools": [
    {
      "version": 6,
      "key": "key"
    },
    {
      "version": 6,
      "key": "key"
    }
  ],
  "skills": [
    {
      "version": 1,
      "key": "key"
    },
    {
      "version": 1,
      "key": "key"
    }
  ],
  "toolKeys": [
    "toolKeys",
    "toolKeys"
  ],
  "classifiers": [
    {
      "instructions": "instructions",
      "criteria": "",
      "type": "bool",
      "key": "key"
    },
    {
      "instructions": "instructions",
      "criteria": "",
      "type": "bool",
      "key": "key"
    }
  ],
  "name": "name",
  "messages": [
    {
      "role": "role",
      "content": "content"
    },
    {
      "role": "role",
      "content": "content"
    }
  ],
  "comment": "comment",
  "model": "{}",
  "state": "archived",
  "outputFormat": {
    "key": ""
  },
  "modelConfigVersion": 1
}
'

import requests

url = "https://app.launchdarkly.com/api/v2/projects/{projectKey}/ai-configs/{configKey}/variations/{variationKey}"

payload = {
    "judgeConfiguration": { "judges": [
            {
                "samplingRate": 0.36160767,
                "judgeConfigKey": "judgeConfigKey"
            },
            {
                "samplingRate": 0.36160767,
                "judgeConfigKey": "judgeConfigKey"
            }
        ] },
    "instructions": "instructions",
    "description": "description",
    "published": True,
    "modelConfigKey": "OpenAI.gpt-4.1",
    "tools": [
        {
            "version": 6,
            "key": "key"
        },
        {
            "version": 6,
            "key": "key"
        }
    ],
    "skills": [
        {
            "version": 1,
            "key": "key"
        },
        {
            "version": 1,
            "key": "key"
        }
    ],
    "toolKeys": ["toolKeys", "toolKeys"],
    "classifiers": [
        {
            "instructions": "instructions",
            "criteria": "",
            "type": "bool",
            "key": "key"
        },
        {
            "instructions": "instructions",
            "criteria": "",
            "type": "bool",
            "key": "key"
        }
    ],
    "name": "name",
    "messages": [
        {
            "role": "role",
            "content": "content"
        },
        {
            "role": "role",
            "content": "content"
        }
    ],
    "comment": "comment",
    "model": "{}",
    "state": "archived",
    "outputFormat": { "key": "" },
    "modelConfigVersion": 1
}
headers = {
    "Authorization": "<api-key>",
    "Content-Type": "application/json"
}

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

print(response.text)

const options = {
  method: 'PATCH',
  headers: {Authorization: '<api-key>', 'Content-Type': 'application/json'},
  body: JSON.stringify({
    judgeConfiguration: {
      judges: [
        {samplingRate: 0.36160767, judgeConfigKey: 'judgeConfigKey'},
        {samplingRate: 0.36160767, judgeConfigKey: 'judgeConfigKey'}
      ]
    },
    instructions: 'instructions',
    description: 'description',
    published: true,
    modelConfigKey: 'OpenAI.gpt-4.1',
    tools: [{version: 6, key: 'key'}, {version: 6, key: 'key'}],
    skills: [{version: 1, key: 'key'}, {version: 1, key: 'key'}],
    toolKeys: ['toolKeys', 'toolKeys'],
    classifiers: [
      {instructions: 'instructions', criteria: '', type: 'bool', key: 'key'},
      {instructions: 'instructions', criteria: '', type: 'bool', key: 'key'}
    ],
    name: 'name',
    messages: [{role: 'role', content: 'content'}, {role: 'role', content: 'content'}],
    comment: 'comment',
    model: '{}',
    state: 'archived',
    outputFormat: {key: ''},
    modelConfigVersion: 1
  })
};

fetch('https://app.launchdarkly.com/api/v2/projects/{projectKey}/ai-configs/{configKey}/variations/{variationKey}', 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://app.launchdarkly.com/api/v2/projects/{projectKey}/ai-configs/{configKey}/variations/{variationKey}",
  CURLOPT_RETURNTRANSFER => true,
  CURLOPT_ENCODING => "",
  CURLOPT_MAXREDIRS => 10,
  CURLOPT_TIMEOUT => 30,
  CURLOPT_HTTP_VERSION => CURL_HTTP_VERSION_1_1,
  CURLOPT_CUSTOMREQUEST => "PATCH",
  CURLOPT_POSTFIELDS => json_encode([
    'judgeConfiguration' => [
        'judges' => [
                [
                                'samplingRate' => 0.36160767,
                                'judgeConfigKey' => 'judgeConfigKey'
                ],
                [
                                'samplingRate' => 0.36160767,
                                'judgeConfigKey' => 'judgeConfigKey'
                ]
        ]
    ],
    'instructions' => 'instructions',
    'description' => 'description',
    'published' => true,
    'modelConfigKey' => 'OpenAI.gpt-4.1',
    'tools' => [
        [
                'version' => 6,
                'key' => 'key'
        ],
        [
                'version' => 6,
                'key' => 'key'
        ]
    ],
    'skills' => [
        [
                'version' => 1,
                'key' => 'key'
        ],
        [
                'version' => 1,
                'key' => 'key'
        ]
    ],
    'toolKeys' => [
        'toolKeys',
        'toolKeys'
    ],
    'classifiers' => [
        [
                'instructions' => 'instructions',
                'criteria' => '',
                'type' => 'bool',
                'key' => 'key'
        ],
        [
                'instructions' => 'instructions',
                'criteria' => '',
                'type' => 'bool',
                'key' => 'key'
        ]
    ],
    'name' => 'name',
    'messages' => [
        [
                'role' => 'role',
                'content' => 'content'
        ],
        [
                'role' => 'role',
                'content' => 'content'
        ]
    ],
    'comment' => 'comment',
    'model' => '{}',
    'state' => 'archived',
    'outputFormat' => [
        'key' => ''
    ],
    'modelConfigVersion' => 1
  ]),
  CURLOPT_HTTPHEADER => [
    "Authorization: <api-key>",
    "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://app.launchdarkly.com/api/v2/projects/{projectKey}/ai-configs/{configKey}/variations/{variationKey}"

	payload := strings.NewReader("{\n  \"judgeConfiguration\": {\n    \"judges\": [\n      {\n        \"samplingRate\": 0.36160767,\n        \"judgeConfigKey\": \"judgeConfigKey\"\n      },\n      {\n        \"samplingRate\": 0.36160767,\n        \"judgeConfigKey\": \"judgeConfigKey\"\n      }\n    ]\n  },\n  \"instructions\": \"instructions\",\n  \"description\": \"description\",\n  \"published\": true,\n  \"modelConfigKey\": \"OpenAI.gpt-4.1\",\n  \"tools\": [\n    {\n      \"version\": 6,\n      \"key\": \"key\"\n    },\n    {\n      \"version\": 6,\n      \"key\": \"key\"\n    }\n  ],\n  \"skills\": [\n    {\n      \"version\": 1,\n      \"key\": \"key\"\n    },\n    {\n      \"version\": 1,\n      \"key\": \"key\"\n    }\n  ],\n  \"toolKeys\": [\n    \"toolKeys\",\n    \"toolKeys\"\n  ],\n  \"classifiers\": [\n    {\n      \"instructions\": \"instructions\",\n      \"criteria\": \"\",\n      \"type\": \"bool\",\n      \"key\": \"key\"\n    },\n    {\n      \"instructions\": \"instructions\",\n      \"criteria\": \"\",\n      \"type\": \"bool\",\n      \"key\": \"key\"\n    }\n  ],\n  \"name\": \"name\",\n  \"messages\": [\n    {\n      \"role\": \"role\",\n      \"content\": \"content\"\n    },\n    {\n      \"role\": \"role\",\n      \"content\": \"content\"\n    }\n  ],\n  \"comment\": \"comment\",\n  \"model\": \"{}\",\n  \"state\": \"archived\",\n  \"outputFormat\": {\n    \"key\": \"\"\n  },\n  \"modelConfigVersion\": 1\n}")

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

	req.Header.Add("Authorization", "<api-key>")
	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.patch("https://app.launchdarkly.com/api/v2/projects/{projectKey}/ai-configs/{configKey}/variations/{variationKey}")
  .header("Authorization", "<api-key>")
  .header("Content-Type", "application/json")
  .body("{\n  \"judgeConfiguration\": {\n    \"judges\": [\n      {\n        \"samplingRate\": 0.36160767,\n        \"judgeConfigKey\": \"judgeConfigKey\"\n      },\n      {\n        \"samplingRate\": 0.36160767,\n        \"judgeConfigKey\": \"judgeConfigKey\"\n      }\n    ]\n  },\n  \"instructions\": \"instructions\",\n  \"description\": \"description\",\n  \"published\": true,\n  \"modelConfigKey\": \"OpenAI.gpt-4.1\",\n  \"tools\": [\n    {\n      \"version\": 6,\n      \"key\": \"key\"\n    },\n    {\n      \"version\": 6,\n      \"key\": \"key\"\n    }\n  ],\n  \"skills\": [\n    {\n      \"version\": 1,\n      \"key\": \"key\"\n    },\n    {\n      \"version\": 1,\n      \"key\": \"key\"\n    }\n  ],\n  \"toolKeys\": [\n    \"toolKeys\",\n    \"toolKeys\"\n  ],\n  \"classifiers\": [\n    {\n      \"instructions\": \"instructions\",\n      \"criteria\": \"\",\n      \"type\": \"bool\",\n      \"key\": \"key\"\n    },\n    {\n      \"instructions\": \"instructions\",\n      \"criteria\": \"\",\n      \"type\": \"bool\",\n      \"key\": \"key\"\n    }\n  ],\n  \"name\": \"name\",\n  \"messages\": [\n    {\n      \"role\": \"role\",\n      \"content\": \"content\"\n    },\n    {\n      \"role\": \"role\",\n      \"content\": \"content\"\n    }\n  ],\n  \"comment\": \"comment\",\n  \"model\": \"{}\",\n  \"state\": \"archived\",\n  \"outputFormat\": {\n    \"key\": \"\"\n  },\n  \"modelConfigVersion\": 1\n}")
  .asString();

require 'uri'
require 'net/http'

url = URI("https://app.launchdarkly.com/api/v2/projects/{projectKey}/ai-configs/{configKey}/variations/{variationKey}")

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

request = Net::HTTP::Patch.new(url)
request["Authorization"] = '<api-key>'
request["Content-Type"] = 'application/json'
request.body = "{\n  \"judgeConfiguration\": {\n    \"judges\": [\n      {\n        \"samplingRate\": 0.36160767,\n        \"judgeConfigKey\": \"judgeConfigKey\"\n      },\n      {\n        \"samplingRate\": 0.36160767,\n        \"judgeConfigKey\": \"judgeConfigKey\"\n      }\n    ]\n  },\n  \"instructions\": \"instructions\",\n  \"description\": \"description\",\n  \"published\": true,\n  \"modelConfigKey\": \"OpenAI.gpt-4.1\",\n  \"tools\": [\n    {\n      \"version\": 6,\n      \"key\": \"key\"\n    },\n    {\n      \"version\": 6,\n      \"key\": \"key\"\n    }\n  ],\n  \"skills\": [\n    {\n      \"version\": 1,\n      \"key\": \"key\"\n    },\n    {\n      \"version\": 1,\n      \"key\": \"key\"\n    }\n  ],\n  \"toolKeys\": [\n    \"toolKeys\",\n    \"toolKeys\"\n  ],\n  \"classifiers\": [\n    {\n      \"instructions\": \"instructions\",\n      \"criteria\": \"\",\n      \"type\": \"bool\",\n      \"key\": \"key\"\n    },\n    {\n      \"instructions\": \"instructions\",\n      \"criteria\": \"\",\n      \"type\": \"bool\",\n      \"key\": \"key\"\n    }\n  ],\n  \"name\": \"name\",\n  \"messages\": [\n    {\n      \"role\": \"role\",\n      \"content\": \"content\"\n    },\n    {\n      \"role\": \"role\",\n      \"content\": \"content\"\n    }\n  ],\n  \"comment\": \"comment\",\n  \"model\": \"{}\",\n  \"state\": \"archived\",\n  \"outputFormat\": {\n    \"key\": \"\"\n  },\n  \"modelConfigVersion\": 1\n}"

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

{
  "judgeConfiguration": {
    "judges": [
      {
        "samplingRate": 0.36160767,
        "judgeConfigKey": "judgeConfigKey"
      },
      {
        "samplingRate": 0.36160767,
        "judgeConfigKey": "judgeConfigKey"
      }
    ]
  },
  "instructions": "instructions",
  "color": "color",
  "_links": {
    "parent": {
      "href": "href",
      "type": "type"
    }
  },
  "description": "description",
  "modelConfigKey": "OpenAI.gpt-4.1",
  "judgingConfigKeys": [
    "judgingConfigKeys",
    "judgingConfigKeys"
  ],
  "version": 5,
  "tools": [
    {
      "customParameters": {
        "key": ""
      },
      "version": 7,
      "key": "key"
    },
    {
      "customParameters": {
        "key": ""
      },
      "version": 7,
      "key": "key"
    }
  ],
  "skills": [
    {
      "version": 9,
      "key": "key"
    },
    {
      "version": 9,
      "key": "key"
    }
  ],
  "createdAt": 1,
  "_publishedAt": 2,
  "_archivedAt": 5,
  "classifiers": [
    {
      "instructions": "instructions",
      "criteria": "",
      "type": "bool",
      "key": "key"
    },
    {
      "instructions": "instructions",
      "criteria": "",
      "type": "bool",
      "key": "key"
    }
  ],
  "name": "name",
  "messages": [
    {
      "role": "role",
      "content": "content"
    },
    {
      "role": "role",
      "content": "content"
    }
  ],
  "comment": "comment",
  "model": "{}",
  "_id": "_id",
  "state": "published",
  "outputFormat": {
    "key": ""
  },
  "key": "key",
  "modelConfigVersion": 6
}

{
  "code": "code",
  "message": "message"
}

{
  "code": "code",
  "message": "message"
}

{
  "code": "code",
  "message": "message"
}

{
  "code": "code",
  "message": "message"
}

Authorizations

Authorization

string

header

required

Path Parameters

projectKey

string

required

The project key

configKey

string

required

The AI Config key

variationKey

string

required

The AI Config variation key

Body

application/json

AI Config variation object to update

classifiers

object[]

Replaces classifier rules on a judge variation. An empty array clears them. When this array is non-empty, messages must be empty.

Show child attributes

comment

string

Human-readable description of what this patch changes

description

string

Description for agent when the config is in agent mode.

instructions

string

Instructions for agent when the config is in agent mode.

messages

object[]

The prompt messages for this variation

Show child attributes

model

object

Freeform JSON object. LaunchDarkly does not validate the keys. In a model config, params and customParams hold parameters that the provider's API accepts, for example {"temperature": 0.7, "maxTokens": 4096}.

modelConfigKey

string

The key that GET /api/v2/projects/{projectKey}/ai-configs/model-configs returns for the model config. Do not use a custom model config's ID, which is the provider model ID.

Example:

"OpenAI.gpt-4.1"

modelConfigVersion

integer

The exact model config version to set this variation to. Requires modelConfigKey to be set (either in this patch or already on the variation), and the version must exist for that model config. When omitted, the variation keeps the version it is currently set to, unless the patch changes modelConfigKey, in which case the variation records the new model config's current (latest) version.

Required range: x >= 1

name

string

A human-readable name for the variation

outputFormat

object

JSON Schema defining the structured output format for the variation.

published

boolean

Whether to publish this version of the variation

state

enum<string>

The state to move the variation to. Omit this field to keep the current state.

archived: Archive a published variation. You cannot archive a variation that is already archived. You also cannot archive the only variation of an AI Config. An AI Config itself cannot be archived.
published: Restore an archived variation. You cannot publish a variation that is already published.

Available options:

archived,

published

Example:

"archived"

tools

object[]

List of tools to use for this variation. The latest version of the tool will be used.

Show child attributes

toolKeys

string[]

List of tool keys to use for this variation. The latest version of the tool will be used.

skills

object[]

List of agent skills to attach to this variation. Replaces the current attachments.

Show child attributes

judgeConfiguration

object

Show child attributes

Example:

{
  "judges": [
    {
      "samplingRate": 0.36160767,
      "judgeConfigKey": "judgeConfigKey"
    },
    {
      "samplingRate": 0.36160767,
      "judgeConfigKey": "judgeConfigKey"
    }
  ]
}

Response

AI Config variation updated

key

string

required

The variation key

_id

string

required

The variation ID

model

object

required

name

string

required

The variation name

createdAt

integer<int64>

required

version

integer

required

The version number of this resource

_links

object

Show child attributes

Example:

{
  "parent": { "href": "href", "type": "type" }
}

color

string

The color used for this variation in the UI

classifiers

object[]

Classifier rules for a judge variation. When this array is non-empty, messages must be empty.

Show child attributes

comment

string

A comment describing this version of the variation

description

string

Returns the description for the agent. This is only returned for agent variations.

instructions

string

Returns the instructions for the agent. This is only returned for agent variations.

messages

object[]

The prompt messages for this variation

Show child attributes

modelConfigKey

string

The key of the model config that this variation uses, not the provider model ID. Absent when the variation does not use a model config.

Example:

"OpenAI.gpt-4.1"

modelConfigVersion

integer

The version of the model config that this variation is pinned to, when it references a custom model config. Absent when the variation does not reference a model config, when the variation is not pinned to a specific version, or when custom model config versioning is not enabled for the project.

outputFormat

object

JSON Schema defining the structured output format for the variation.

state

enum<string>

The state of the variation.

archived: The variation is archived. To restore it, update the variation with state set to published.
published: The variation is active.

Available options:

archived,

published

Example:

"published"

_archivedAt

integer<int64>

_publishedAt

integer<int64>

tools

object[]

The tools this variation can use

Show child attributes

skills

object[]

The agent skills this variation can use

Show child attributes

judgeConfiguration

object

Show child attributes

Example:

{
  "judges": [
    {
      "samplingRate": 0.36160767,
      "judgeConfigKey": "judgeConfigKey"
    },
    {
      "samplingRate": 0.36160767,
      "judgeConfigKey": "judgeConfigKey"
    }
  ]
}

judgingConfigKeys

string[]

Keys of the judge AI Configs attached to this variation