Grok Build with Nbility
Grok Build and the community Grok CLI are different projects. This page covers versions with custom TOML model configuration; follow current xAI release documentation for installation and schema details.
Register a model
After confirming support for [model] configuration, edit user-level ~/.grok/config.toml:
[models]
default = "YOUR_MODEL_ID"
[model."YOUR_MODEL_ID"]
base_url = "https://api.nbility.ai/v1"
env_key = "NBILITY_API_KEY"
api_backend = "chat_completions"
context_window = 8192
max_completion_tokens = 1024
Replace both ID occurrences with the same real model ID. Keep quotes around names containing dots; do not substitute a display nickname. Adjust the illustrative context and output limits to your model.
Select and verify
Set export NBILITY_API_KEY="YOUR_API_KEY", launch Grok Build, and select the configured model. Test a short message and verify it in Usage logs.
If the installed version does not expose custom models or ignores this file, verify which Grok project you installed before blaming the API token. Use its matching configuration guide, or choose OpenCode with a documented custom provider flow.
References
Adapted and rewritten for Nbility from aicentos’s Grok Build guide, licensed under CC BY 4.0. Its account, plan, and proprietary service instructions do not apply here. See the client’s official documentation.