Using Grok Build in Warp with a SuperGrok or X Premium Subscription

On June 15, 2026, xAI added a practical shortcut for developers already paying for Grok: connect that existing Grok or X Premium subscription inside Warp, pick grok-build-0.1, and use it from Warp’s agent interface. The announcement is small, but the workflow shift is real. Instead of treating Grok Build as a separate terminal CLI, Warp users can put xAI’s coding model into the same agent surface where they already run commands, inspect diffs, and manage coding sessions.

xAI says Warp is “used by almost one million developers” and that existing subscriptions now unlock Grok models in Warp, including grok-build-0.1, the model behind the Grok Build CLI (xAI). Warp’s own docs describe Warp as an open-source Agentic Development Environment that combines a Rust terminal, local agents, cloud agents, code review, model choice, and Oz orchestration (Warp docs).

What Changed on June 15

The new integration is authentication glue plus model placement. xAI’s setup instructions are only three steps: download Warp, open Agent > Warp Agent, click Connect SuperGrok subscription, then switch to grok-build-0.1 and start prompting (xAI).

That matters because agentic coding tools have been splitting into two camps:

1. Standalone CLIs that own the terminal session.
2. IDE or terminal environments that orchestrate models, tools, files, commands, and review loops.

Grok Build started in the first camp. xAI introduced it as an early beta terminal coding agent for SuperGrok and X Premium Plus subscribers on May 25, 2026, with a curl installer and a plan-review-approve workflow (xAI). The Warp integration moves the same model into a broader terminal workspace.

This is not a new model release. It is a distribution and workflow release. For developers, that is still meaningful. The fewer context switches between “terminal,” “agent,” “diff review,” and “model picker,” the more likely the agent becomes part of daily work instead of a side experiment.

Eligibility and Setup

The eligibility headline from xAI is direct: “your Grok or X Premium subscription” can be used inside Warp (xAI). The setup button in the announcement is labeled Connect SuperGrok subscription, so expect the flow to verify your paid Grok/X account rather than ask for an xAI API key.

xAI’s public pricing page lists SuperGrok at $30/month, with higher rate limits, Grok 4 access, connectors, Expert, SOC 2 compliance, and image and video generation (xAI pricing). X Premium pricing is more variable. X’s help center says Premium pricing is localized and starts at $3/month or $32/year, with prices shown in the web or mobile purchase flow for each region and platform (X Help). If your goal is only Grok Build in Warp, check the exact plan shown in your account before upgrading.

The setup path should look like this:

Warp → Settings → Agent → Warp Agent
→ Connect SuperGrok subscription
→ Model picker → grok-build-0.1
→ Prompt from a repo-backed terminal session

The key operational detail: this is subscription-based access inside Warp, not the same thing as calling the xAI API directly with XAI_API_KEY. If you want API billing, rate-limit control, or server-side automation outside Warp, the API path is separate.

The Model Numbers Developers Should Care About

xAI made grok-build-0.1 available through the xAI API in public beta on May 29, 2026, calling it its “fastest coding model” and saying it is trained for agentic coding tasks including web development, debugging, and MCP support (xAI). The same post says the model serves at 100+ tokens/second and costs $1 per million input tokens and $2 per million output tokens on the API.

The model docs add the context window, modalities, cached-token price, aliases, and rate limits (xAI docs):

Item	Verified value
Model name	grok-build-0.1
Context window	256,000 tokens
Modalities	Text and image input, text output
API price	$1.00 / 1M input tokens, $2.00 / 1M output tokens
Cached input	$0.20 / 1M cached tokens
Claimed serving speed	100+ tokens/second
Listed aliases	grok-code-fast-1, grok-code-fast, grok-code-fast-1-0825
Region in docs	us-east-1
Rate limits in docs	1,800 RPM, 10,000,000 TPM

Do not overread these numbers. 100+ tokens/second is a vendor throughput claim, not a coding benchmark. I did not find a primary xAI SWE-bench, Terminal-Bench, or similar public benchmark score for grok-build-0.1 in the sources above. For now, the defensible claim is narrower: this is a fast, low-cost coding-oriented model with a large context window and first-party support in Grok Build plus Warp.

Why Warp Is a Natural Host

Warp is no longer just a polished terminal. Its docs describe terminal mode, agent mode, a file tree, code editor, LSP-backed editing, interactive code review, third-party CLI agents, cloud agents, model choice, and task lists (Warp docs). That is exactly the surface area where a coding model becomes useful.

A standalone agent CLI can edit files and run commands, but it often lives beside your real workflow. Warp puts the agent where the work already happens: shell output, failed test logs, file context, prompt history, command blocks, and generated diffs.

Warp’s agent docs list capabilities that map cleanly to coding-agent loops: attach files, URLs, images, code blocks, and selections as context; allow full terminal use; review generated diffs; track task lists; and use web search for current information (Warp agents). Its model-choice docs also say Warp supports a curated set of models from OpenAI, Anthropic, Google, xAI, and hosted open-source providers, with model fallback and per-profile defaults (Warp model choice).

That makes grok-build-0.1 less of a separate tool and more of another runtime target. You can use it for fast edits, debugging, and scaffolding, then switch models when you want a different reasoning style or provider policy.

What This Means for Daily Coding Work

The most obvious win is for developers who already pay for Grok or X Premium and already use Warp. They now get another coding model in the same place they run local commands. No separate CLI install is required for this Warp path. No API key setup is mentioned in xAI’s Warp announcement. That lowers the activation cost.

The second win is model routing by task. Use grok-build-0.1 when speed and coding orientation matter: generate a migration script, chase a failing test, refactor a route handler, wire a small MCP tool, or ask the agent to inspect terminal output and propose the next command. Use other Warp models when you want a different tradeoff.

The third win is review discipline. Grok Build’s own CLI pitch emphasizes plan, review, and approve. Warp’s interface already has interactive code review and task tracking. That combination is healthier than “agent runs wild in a shell.” For serious repos, the workflow should be:

1. Ask for a plan.
2. Review scope.
3. Let the agent edit.
4. Inspect the diff.
5. Run targeted tests.
6. Commit only after human review.

The risk is subscription ambiguity. xAI says Grok or X Premium subscriptions work, but plan names across Grok, SuperGrok, X Premium, and X Premium+ have shifted over time. If your account does not show grok-build-0.1 in Warp, the first debugging step is not your repo. It is billing and entitlement.

Bottom Line

This is a distribution move, not a model breakthrough. Still, it is the kind of move that changes habits. grok-build-0.1 already had a CLI and API; now it also has a seat inside Warp’s agentic terminal environment.

For Warp users, the practical question is simple: if you already have a qualifying Grok or X Premium subscription, try grok-build-0.1 on a contained repo task and judge it by diff quality, command safety, and how often you need to correct it. Speed and price are nice. A clean, reviewable patch is what matters.

Readers who want to try these models hands-on outside Warp can call Claude and other models on onehop with an OpenAI-compatible API by changing one base_url: call Claude and other models on onehop. onehop is cheaper than first-party providers, and new accounts get $10 free credit with no card required: sign up for $10 free credit.