Which API should you use with Grok Build?

For Grok Build with more than Grok models, the best API is a cost-optimized custom model: add [model.standardcompute] to ~/.grok/config.toml with base_url https://api.stdcmpt.com/v1, api_backend = "chat_completions" and env_key = "STANDARDCOMPUTE_API_KEY", and Standard Compute's router sends each step to the cheapest model that can do it (Claude Fable 5.1 or GPT-6 Astra only when it counts), on a plan from $19 a month with a $20 compute budget. Signing in with an xAI account runs Grok's own models; The X Subscription's paid tiers ($8 to $200 a month) include a weekly pool for first-party Grok usage, and an XAI_API_KEY bills per token. Grok Build itself is open source (Apache 2.0).

Light or occasional use

If Grok models are all you want and your work fits the weekly pool, an X Subscription plan you already pay for is the simplest bill. For occasional headless runs, an xAI API key billed per token will likely cost less.

Daily use, cost-optimized

When Grok Build runs long sessions with parallel subagents and you want Claude, GPT or open models next to Grok, a Standard Compute custom model sends each step to the cheapest model that can do it, and for every model we vet providers for price, uptime and speed, with automatic fallback. Opt-in Europe or US Compliance rules apply per key, team or company, and we don't train on your prompts or store them. The bill stays at the monthly plan price: requests stop with HTTP 402 when the compute budget is used, with nothing billed on top.

API and subscription options for Grok Build

Prices and allowances can change. Grok Build's custom model fields, credential order and environment variables come from xAI's Grok Build overview, settings reference and enterprise docs; The X Subscription prices from xAI's pricing doc, all read on October 15, 2026. Grok Build: getting started · Grok Build: settings reference · The X Subscription pricing · Grok Build repo. Spot an error? Tell us.

Add Standard Compute as a Grok Build custom model

[model.standardcompute]
model = "standardcompute"
base_url = "https://api.stdcmpt.com/v1"
name = "Standard Compute (routed)"
env_key = "STANDARDCOMPUTE_API_KEY"
api_backend = "chat_completions"
context_window = 200000

[model.sc-sonnet]
model = "anthropic/claude-sonnet-5.5"
base_url = "https://api.stdcmpt.com/v1"
name = "Claude Sonnet 5.5 (Standard Compute)"
env_key = "STANDARDCOMPUTE_API_KEY"
api_backend = "chat_completions"

[models]
default = "standardcompute"

Export STANDARDCOMPUTE_API_KEY, run grok inspect to check the config loaded, then start grok. Switch with /model sc-sonnet in the TUI or grok -m sc-sonnet. On Windows the file is %USERPROFILE%\.grok\config.toml.

Grok Build questions

Do I have to log out of my xAI account to use a custom model in Grok Build?

No, not when the model has its own key. Grok Build resolves credentials per model in this order: model.api_key, then model.env_key, then the active session token, then XAI_API_KEY. A [model.standardcompute] entry with env_key = "STANDARDCOMPUTE_API_KEY" therefore uses your Standard Compute key even while you are signed in, and Grok models keep using your xAI login.

Which api_backend should I use for a gateway in Grok Build?

The protocol the gateway serves. Custom models accept chat_completions, responses or messages (OpenAI-compatible or Anthropic Messages). Standard Compute serves Chat Completions and Responses at https://api.stdcmpt.com/v1; chat_completions is the safe default. GROK_MODELS_BASE_URL is a different mechanism: it points Grok Build's whole model list at {base}/models instead of adding one model.

Does Grok Build's web search work on a custom model?

The web_search tool runs on its own model, set with [models] web_search or GROK_WEB_SEARCH_MODEL, separate from your chat model. supports_backend_search marks whether an endpoint supports Grok-hosted server-side search tools; leave it off for a third-party endpoint such as Standard Compute.