Grok Build From $30/mo, or grok-build-0.1 API at $1/M (2026)

30-second answer. Grok Build, xAI’s terminal coding agent, launched May 14, 2026 limited to SuperGrok Heavy ($300/month), but on May 24 xAI opened it to all SuperGrok ($30/month) and X Premium+ ($40/month) subscribers. So the agent now starts at $30/month, which is cheap for a first-party terminal coding agent. The same intelligence is also a model called grok-build-0.1, sold token-by-token on the xAI API at $1.00/M input, $2.00/M output (cached input $0.20/M, 256K context). If you want an even cheaper Grok, Grok 4.1 Fast is $0.20/M input and $0.50/M output. So the real question is no longer “is $300 worth it” but “do I want a finished agent on a flat $30 sub, or metered tokens with my own CLI.”

$30 buys the polished agent on a subscription. $1.00 per million input tokens buys the same model à la carte with your own CLI. Decide which one fits how you actually work.

TL;DR: Subscription or API?

The split is clean. Subscription is a flat fee for the finished app. API is metered access to the same intelligence with no app, but no floor either, and it drops into your own tooling.

What “$30/mo” Actually Buys

Grok Build went into beta on May 14, 2026 as a terminal-native coding agent with a Rust CLI, in the same lane as Claude Code and Codex CLI. At launch it shipped only to SuperGrok Heavy subscribers. That changed fast: on May 24, 2026, xAI expanded access to all SuperGrok ($30/month) and X Premium+ ($40/month) subscribers. Here is where the agent now sits on xAI’s consumer ladder:

So “you need $300 for Grok Build” is now wrong. The cheapest plan that includes the Grok Build agent is $30 SuperGrok. That is genuinely cheap for a polished, first-party terminal coding agent, in the same ballpark as Claude Code and Codex CLI subscriptions.

The $300 SuperGrok Heavy tier still exists, but its pitch is no longer “the only way to get the agent.” It is the heavy-user tier: Grok 4 Heavy multi-agent reasoning, the highest rate limits, and priority during peak load.

The Model Under the Agent: grok-build-0.1

The Grok Build agent is a wrapper. The intelligence is a model named grok-build-0.1, and xAI sells that model directly on its API, metered:

Those are the numbers on xAI’s own model docs as of late June 2026. The model is roughly 20% cheaper per token than the Grok 4.3 flagship ($1.25/$2.50), and it supports prompt caching, which drops repeated context to $0.20/M. For an agent that re-sends a large system prompt and the same files on every turn, caching is where the real savings live.

Pair that API key with any open-source agent CLI that takes a custom OpenAI-compatible endpoint and you get most of the Grok Build experience for the cost of the tokens you actually spend, with no subscription at all.

Pricing Math: When Does the Sub Break Even?

Let’s do the arithmetic instead of guessing. Take grok-build-0.1 at $1.00/M input and $2.00/M output, and assume a typical agentic coding turn: a chunky prompt plus context in, a moderate diff out. Caching matters a lot here, so I’ll show both with and without it.

Cost per “turn” with a rough model: assume ~30K input tokens (instructions plus a few files) and ~3K output tokens per agent turn.

• Input: 30,000 × $1.00 / 1,000,000 = $0.030
• Output: 3,000 × $2.00 / 1,000,000 = $0.006
• Per turn ≈ $0.036 (no caching)

With prompt caching on the repeated system prompt and stable files, most of that 30K input drops to the $0.20/M cached rate. Say 25K cached + 5K fresh:

• Cached input: 25,000 × $0.20 / 1,000,000 = $0.005
• Fresh input: 5,000 × $1.00 / 1,000,000 = $0.005
• Output: 3,000 × $2.00 / 1,000,000 = $0.006
• Per turn ≈ $0.016 (with caching)

Now the break-even against the flat subscriptions that include the agent:

Read that table slowly. To get $30 of metered value out of grok-build-0.1 with caching on, you need around 85 agent turns per working day. A developer who actually uses Grok Build most days clears that easily, so the $30 subscription is usually the better deal for steady daily use. The API wins for the lighter cases: occasional use, people who don’t want a subscription, or anyone who wants to wire the model into their own CLI. Below roughly 85 turns/day, metered billing comes out cheaper.

This flips the old assumption. When the only plan was $300, almost everyone was better off metered. Now that the agent starts at $30, the math tips the other way for regular users.

If you want to see this logic applied across providers rather than just Grok, the companion piece on how to reduce AI API costs walks through caching, model tiering, and routing in more depth.

The Even-Cheaper Path: Grok 4.1 Fast

grok-build-0.1 is tuned for agentic coding, but it is not the cheapest Grok. If your workload is high-volume, long-context, or not strictly coding-agent-shaped, Grok 4.1 Fast undercuts it hard:

Grok 4.1 Fast is five times cheaper on input than grok-build-0.1 and carries a 2M-token context window. The headline “$0.20/M” in this article is that input rate. It is the floor price for touching a Grok model at all in 2026. The tradeoff: it is a general fast model, not the coding-specialized build model, so for heavy refactoring agents you may prefer grok-build-0.1’s tuning. For batch summarization, retrieval over big documents, or cost-sensitive pipelines, Grok 4.1 Fast is the obvious call.

For the full per-model rundown, including exact model IDs and a working setup, see the existing Grok API pricing, setup and access guide. This post is about the subscription-versus-API decision specifically; that one is the reference for getting any Grok call working in the first place.

How to Run a Grok Model Pay-As-You-Go

The xAI endpoint is OpenAI-compatible, so this is a two-line change in any existing OpenAI SDK setup. Below, the literal grok-build-0.1 path goes straight to xAI (it is the only place that model lives today); the general Grok pay-as-you-go path can also route through an aggregator.

Python: call grok-build-0.1 on the xAI API

from openai import OpenAI

client = OpenAI(api_key="XAI_API_KEY", base_url="https://api.x.ai/v1")

resp = client.chat.completions.create(
    model="grok-build-0.1",
    messages=[{"role": "user", "content": "Refactor this function for readability."}],
)
print(resp.choices[0].message.content)

Node, same shape

import OpenAI from "openai";

const client = new OpenAI({ apiKey: "XAI_API_KEY", baseURL: "https://api.x.ai/v1" });

const resp = await client.chat.completions.create({
  model: "grok-build-0.1",
  messages: [{ role: "user", content: "Write tests for this module." }],
});
console.log(resp.choices[0].message.content);

Drop that key into an open-source agent CLI (Cline, Aider, OpenCode, and similar all accept a custom base URL and model name) and you have a Grok coding agent that bills per token instead of per month.

One key across providers

If you want Grok next to Claude, GPT, and Gemini behind a single key and one bill, an aggregator gateway is the simplest route. Note one honesty check: as of June 2026, ofox.ai carries xAI’s flagship Grok tiers but not grok-build-0.1 specifically, so for the build model you hit xAI directly. For general Grok pay-as-you-go and multi-provider routing, ofox.ai’s multi-model setup keeps the same OpenAI-compatible shape:

client = OpenAI(api_key="OFOX_API_KEY", base_url="https://api.ofox.ai/v1")
# swap the model string to route across providers, one key

That’s the entire CTA: one endpoint, swap the model string, pay for what you use. No subscription floor unless you choose it.

When the Subscription Still Wins

This is not a “never subscribe” article. There are real cases where a flat plan is the rational buy:

• You use the agent steadily. Past roughly 85 turns/day, the $30 SuperGrok plan is cheaper than metered grok-build-0.1, and it caps your spend with no bill anxiety.
• You want the first-party agent. A polished, supported Grok Build install with zero glue code has real value if your time is expensive, and $30 is a low price for it.
• You want Grok 4 Heavy. The multi-agent reasoning model with parallel test-time compute is a SuperGrok Heavy ($300) inclusion, not an API line item you can casually meter the same way.
• You want max rate limits and priority during peak load. Heavy users hitting throttles on lower tiers pay the $300 tier for headroom.

The mistake now is assuming you must pay $300 for the agent at all. You don’t. Start at $30, and only step up to Heavy if you specifically need Grok 4 Heavy or the top rate limits. If your usage is light, skip the sub and meter the model instead.

When NOT to Use Either (and What to Use Instead)

If you’re not committed to Grok at all, don’t anchor on it just because Grok Build made headlines. For pure coding-agent work, Claude Code and Codex CLI are bundled into $20/month plans, right alongside $30 SuperGrok, and many developers find their coding quality competitive. If your priority is cheapest-capable tokens regardless of brand, compare Grok 4.1 Fast against other budget tiers in the AI API pricing comparison before locking in. Grok’s edges are real-time X search and the 2M-context Fast tier; if you don’t need those, the decision opens back up.

Alternatives at a Glance

• SuperGrok ($30/mo). Cheapest plan that includes the first-party Grok Build agent. Best for steady daily use.
• xAI API direct. The only home of grok-build-0.1 today. Pay-as-you-go, OpenAI-compatible, no floor. Best for light use or your own CLI.
• ofox.ai. One OpenAI-compatible key across Grok flagship tiers, Claude, GPT, Gemini and more. Good when you route across providers (note: not grok-build-0.1 specifically).
• SuperGrok Heavy ($300/mo). Grok 4 Heavy plus max rate limits plus priority. The heavy-user tier, not a requirement for the agent.
• Claude Code / Codex CLI. Coding agents bundled at ~$20/mo if you’re not married to Grok.

Sources Checked for This Refresh

• xAI Docs, grok-build-0.1 model page (verified 2026-06-29): https://docs.x.ai/developers/models/grok-build-0.1
• xAI News, Grok Build CLI access expansion (verified 2026-06-29): https://x.ai/news/grok-build-cli
• xAI Docs, Grok 4.3 model page (verified 2026-06-29): https://docs.x.ai/developers/models/grok-4.3
• xAI Docs, models index (verified 2026-06-29): https://docs.x.ai/developers/models
• OpenRouter, Grok 4.3 pricing and specs (verified 2026-06-29): https://openrouter.ai/x-ai/grok-4.3
• Engadget, xAI’s coding agent Grok Build (verified 2026-06-29): https://www.engadget.com/2173482/xai-coding-agent-grok-build/
• Fello AI, Grok subscription tier pricing (verified 2026-06-29): https://felloai.com/grok-pricing/
• ofox.ai models page snapshot (verified 2026-06-29)