Grok Bot for Coding: Can Grok Bot Code?

1
Open ReasonOn the web, or in the Reason app for Mac.
2
Connect GitHubChoose the repositories it works in.
3
Describe the taskReason does the work and opens a pull request.

Reason works on its own cloud computer, on a Reason plan.

Can Grok Bot code? Yes. Here is where it fits.

Yes, Grok Bot can code: each Bot has a persistent cloud computer with a terminal, browser and filesystem, writes and runs code there, and xAI's own engineering guide has Bots start Cursor cloud agents for bigger changes and review their pull requests. If you want that always-on teammate to be a frontier coding agent itself, Reason has its own computer, memory, schedules and Slack, works with Claude Code and Codex, and opens the pull request.

By The Reason Machines teamUpdated Nov 2026

How to use a Grok Bot-style teammate for coding with Reason

Open Reason on the web or the Mac app
Sign in at reasonmachines.com, or download the Reason app for macOS from reasonmachines.com/download. Both open the same Sessions. Sending work to Reason needs a Pro or Max plan.
Connect GitHub
In Customize, open GitHub and choose Connect GitHub, then pick the repositories Reason may work in. Reason opens pull requests from its own branches; your branch protection and required checks still apply.
Describe the task in the composer
Name the repository and say what you want, for example "fix the failing checkout test and open a pull request". Reason works on its own cloud computer, runs the tests, and opens the pull request for you to review.
Keep it working when you're away
Mention Reason in Slack to hand it a task from a thread, and set up a Loop for work that repeats, such as a nightly dependency check or a morning CI sweep. Reason keeps memory across Sessions, so it remembers how your repository works.

What Grok Bot can do with code

xAI describes Bots as AI teammates that each work on a persistent cloud computer with a browser, command line and filesystem, and that keep working while your laptop is closed. A Bot can run commands there, add MCP servers, and connect GitHub: xAI's October 22 changelog says the GitHub connector can list repositories, compare commits, list a pull request's commits and read comments.

For larger coding work, xAI's engineering guide says every Bot can create Cursor cloud agents, read their transcripts, review proofs attached to pull requests and send follow-ups. Grok Bot picks the model per task, with no model picker, and runs on Cursor's cloud; it's included on the Premium ($30), Super ($100) and Ultra ($200) X Subscription plans and draws from one weekly pool shared with Grok and Cursor.

Is Grok good at coding? The public numbers

Benchmark	Result	Where it comes from
Terminal-Bench 4.0	Grok 4.7 (xHigh) 37.6%; Claude Fable 5.1 (Max) 57.9%; GPT-5.6 Sol (Max) 37.3%	xAI's Grok 4.7 launch table
DeepSWE v1.1	Grok 4.7 71.0% (high effort); GPT-5.6 Sol 72.7%; Claude Fable 5.1 70.0%	xAI's Grok 4.7 launch table
Coding Agent Index v2.0	Claude Code on Opus 5.5: 62; Codex on GPT-6 Astra: 55; Grok Build on Grok 4.7: 47	Artificial Analysis comparison pages

Numbers as published, checked Nov 2, 2026. These measure models and coding harnesses, not Grok Bot itself, which nobody benchmarks as a coding agent: Grok Build is xAI's coding agent, and Grok Bot routes work to the model it picks. Artificial Analysis's v2.0 index (October 2026) combines FrontierCode v1.1, Terminal-Bench 4.0 and SWE-Atlas-QnA-AA.

Reason's own result is on the earlier v1.5 index: on GPT-6 Astra, Reason 1.1 scored 65.1 against Codex's 62.5 on the same tasks and graders, at 22% lower cost per solved task. It's Reason's benchmark of its own harness, published with a reproducible runner, and not directly comparable with v2.0 scores.

Grok Bot or a coding agent? What each is built for

	Grok Bot	Reason
What it is	Always-on AI teammates for any job, from xAI	An always-on teammate that is also a frontier coding agent
Computer	One persistent cloud computer per user, shared by your Bots	Its own cloud computer per Session, or your own computer through the Mac app
Memory and schedules	Memory per Bot; routines on a schedule or on events	Memory across Sessions; Loops on a schedule, a webhook, GitHub events or Slack messages
Where you talk to it	Grok Bot apps, X Chat, @bot on X, a shared Bot in Slack	Web app, Mac app, iOS, Slack, and the command line
Models	Grok Bot picks the model; no model picker	Pick per task: Reason's models or the Claude, ChatGPT, Grok, Copilot and other accounts you connect
Coding output	Hands larger changes to Cursor cloud agents and reviews their pull requests	Writes the change itself, runs the tests, checks the app in a browser and opens the pull request
Pricing	X Subscription Premium $30, Super $100, Ultra $200 a month	Pro $50, Max $200 a month; credits for Reason's models and cloud computers

You don't have to pick one. Add the Reason MCP server (https://mcp.reasonmachines.com/mcp) to Grok Bot and it can start Reason sessions for coding work, or message the Claude Code or Codex session on your own computer over Reason Network. Hit your limit? Reason keeps going on your other connected accounts or Reason credits.

Sources (checked Nov 2, 2026)

Questions

Can Grok Bot code?

Yes. Each Bot has a persistent cloud computer with a terminal, browser and filesystem, and can write and run code there. For bigger changes, xAI's engineering guide has Bots create Cursor cloud agents and review the pull requests they open.

Is Grok Bot a coding agent?

It's a general always-on teammate that can code, not a dedicated coding agent. xAI's coding agent is Grok Build, a terminal and headless agent. Grok Bot can start Cursor cloud agents or, with the Reason MCP server, Reason sessions for coding work.

How do I use Grok Bot for coding?

Give a Bot your GitHub connection and a clear task, or have it start a Cursor cloud agent. To have a coding agent do the work with its own computer and open the pull request, add the Reason MCP server to Grok Bot, or give the task to Reason directly on the web or in the Mac app.

Is Grok good at coding?

On xAI's own launch table, Grok 4.7 (xHigh) scores 37.6% on Terminal-Bench 4.0 against 57.9% for Claude Fable 5.1 (Max), and 71.0% on DeepSWE v1.1 (high effort) against 72.7% for GPT-5.6 Sol. On Artificial Analysis's Coding Agent Index v2.0, Grok Build on Grok 4.7 scores 47 and Claude Code on Opus 5.5 scores 62.

Can you vibe code with Grok Bot?

You can describe an app in plain words and have a Bot build it on its computer, on Premium or a higher plan. For a change to a real repository that should end in a reviewed pull request, a coding agent such as Reason, Claude Code or Codex is the better fit.

Is Grok Bot good for programmers?

It's useful for the work around code: watching events with routines, reproducing a bug, reviewing pull requests and delegating. Programmers who want the agent that writes the change to choose its model, run on its own computer per task and open the pull request use a coding agent next to it.

What happens when I hit a usage limit in Reason?

Hit your limit? Reason keeps going on your other connected accounts or Reason credits. When the account a turn runs on hits its limit, Reason moves the turn to the same model on your other connected accounts, then to Reason credits if your workspace can pay, and says so in the conversation.

Is Reason made by xAI?

No. Grok Bot is a product of xAI. Reason is built by Reason Machines and is not affiliated with or endorsed by xAI, Anthropic, OpenAI or Cognition. Facts about other products on this page come from their own pages, linked under Sources.