Models in SourceCraft Code Assistant
Warning
This feature is only available in Visual Studio Code.
SourceCraft models
In Code Assistant, you can select SourceCraft AI models available through your organization. The SourceCraft Default model is an ensemble of models: the service automatically selects a model for the task.
To select a different model, you need a paid subscription within the organization selected for deducting neurocredits. The model must be approved by the organization's Administrator or Owner. Open the list of models in the chat and select the one you need. For more information, see Configuring AI models in SourceCraft.
Using SourceCraft models is paid for in neurocredits. For internal models, gift neurocredits, neurocredits from the service plan, and additional paid usage are utilized. For external models, only the neurocredits included in the service plan and additional paid usage are available; you cannot use gift neurocredits for them.
For SourceCraft models, you can enable or disable reasoning and adjust the reasoning effort in the plugin settings. These settings affect token consumption.
Configuration profiles
Model configuration profiles allow you to create and switch between different Code Assistant sets of settings, including connecting AI providers using your own API keys.
Having multiple configuration profiles allows you to quickly switch between different AI providers, models, and settings without having to reconfigure each time you want to modify your setup.
Configuration profiles may include:
- AI providers (Yandex Cloud, OpenAI, Anthropic, OpenRouter, Glama, etc.).
- API keys and authentication data.
- Selecting a model from the connected provider
- Temperature settings to control response randomness.
- Computing budgets.
- Provider-specific settings.
- Configuration editing via diffs.
- Rate limit settings.
Note that available settings vary depending on your provider and model. Each provider offers different configuration options, and even within a single provider, different models may support different ranges of options or features.
Creating and managing profiles
Creating a profile
- In the chat's top panel, click Settings.
- Click next to the Profile selector.
- Enter a name for the new profile and click Create Profile.
- Configure the profile settings:
-
Select an API Provider.
-
Enter the API Key.
Note
API keys are securely stored in VSCode Secret Storage and are never exposed in plain text.
-
Select a Model.
-
Set the Rate limit for this profile. The default value is 0 (disabled), which is suitable for most users. If needed, you can set a minimum time (in seconds) between API requests for a specific profile to manage costs or avoid provider limits.
-
Configure the model settings, such as temperature (Use custom temperature).
-
Switching between profiles
You can switch between profiles in the following ways:
-
In the Code Assistant settings:
- In the chat's top panel, click Settings.
- Under Profile, select the required profile.
-
In the chat's bottom menu, open the drop-down list of model configuration profiles and select the profile.
Pinning and sorting profiles
The drop-down list of model configuration profiles supports pinning profiles for quick access:
-
Hover over any profile in the drop-down list.
-
To add the profile to your pinned list, click .
You can unpin the profile by clicking again.
Pinned profiles are displayed at the top of the drop-down list, sorted alphabetically.
Unpinned profiles are displayed below the divider, also sorted alphabetically.
Editing and deleting profiles
-
In the chat's top panel, click Settings.
-
Under Profile, select the required profile.
-
Edit your profile settings.
To delete a profile, click next to it.
To rename a profile, click next to it.
Note
You cannot delete a single profile.
Linking profiles to modes
In the chat's top panel, click and select Modes. You can explicitly link a specific configuration profile with a specific mode. This system also automatically memorizes which profile you last used with each mode, streamlining your workflow.