Use i14 Models in Cline and Kilo Code

Cline and Kilo Code are AI coding agents for VS Code. Both can talk to any OpenAI-compatible endpoint, so you can point them at the i14 API and use Sugar Glider (i14-sugarglider-nemotron3-super-120B-v1) for chat, file edits, and tool use. Pick your extension below.

Before you start, create an API key at platform.i14.ai.

Cline

1. Install Cline

   Install the Cline extension from the VS Code Extensions view, then open Cline from the Activity Bar.

2. Choose the OpenAI Compatible provider

   Open Cline’s Settings and set API Provider to OpenAI Compatible.

3. Enter the i14 connection details

   Fill in these fields:

   Field	Value
   Base URL	https://api.i14.ai/v1
   API Key	Your i14 API key
   Model ID	i14-sugarglider-nemotron3-super-120B-v1

4. Set the token limit

   Under Model Configuration, set Context Window Size to 32768.

   Caution

   Don’t skip this step. Without this setting, Cline sends no max_tokens, so the i14 API stops each reply after 100 tokens. Cline then fails longer tasks with “Model reached the maximum output token limit before completing the turn.” With Context Window Size set, Cline sends a max_tokens value and completes the task.

5. Click Done and start a task

   Click Done, then send Cline a task. Sugar Glider can read files and run tools in Cline.

Kilo Code

1. Install Kilo Code

   Install the Kilo Code extension from the VS Code Extensions view, then open Kilo Code from the Activity Bar.

2. Add a custom provider

   Open Kilo Code’s settings, go to Providers, and choose Custom provider (“Add a custom provider by base URL”).

3. Enter the i14 connection details

   Fill in the form:

   Field	Value
   Provider ID	i14
   Display name	i14
   Provider API	Leave the default (@ai-sdk/openai-compatible)
   Base URL	https://api.i14.ai/v1
   API key	Your i14 API key
   Models → ID	i14-sugarglider-nemotron3-super-120B-v1
   Models → Name	Sugar Glider

   Click Submit.

4. Add the token limits

   The form saves the provider entry to your global Kilo config file, but it does not set token limits. To open the file, click Global Config in the header of Kilo Code’s settings. The file is ~/.config/kilo/kilo.jsonc (on Windows, C:\Users\<username>\.config\kilo\kilo.jsonc).

   Add a limit block to the model entry, with context set to 32768 and output set to 16384. The entry then looks like this. Kilo Code stores the API key separately in its credential store, not in this file.

   {
     "provider": {
       "i14": {
         "npm": "@ai-sdk/openai-compatible",
         "name": "i14",
         "options": {
           "baseURL": "https://api.i14.ai/v1"
         },
         "models": {
           "i14-sugarglider-nemotron3-super-120B-v1": {
             "name": "Sugar Glider",
             "limit": {
               "context": 32768,
               "output": 16384
             }
           }
         }
       }
     }
   }

   Caution

   Don’t skip this step. Without the limit block, Kilo Code sends max_tokens: 32000, and requests can fail.

5. Select the model

   Select Sugar Glider from the i14 provider as your model, then start a task. Sugar Glider can read files and run tools in Kilo Code.

Note

With this configuration, Kilo Code sends limit.output as max_tokens on every request, so replies are not cut off at the API’s 100-token default.

Troubleshooting

• 403 with model_not_allowed - check that the model ID is exactly i14-sugarglider-nemotron3-super-120B-v1.
• 401 with invalid_api_key - re-enter your API key. See API Key and Authentication for the full list of error codes.