Grok Bot iOS + Which AI Models Should Write And Check Your iPhone Tasks, With Real GoldieBench Scores

Grok Bot iOS has no model menu, so the GoldieBench question is which models you pair with it: one to write the brief you send from your iPhone, and one to check what comes back before you tap Allow.

The Grok Bot FAQ says Grok Bot picks the backend model for each task, there is no model picker, and the choice can differ from one request to the next.

Every GoldieBench score below was read from the live goldiebench.com pages on 2 November 2026, and each one names the page it came from.

If you want my team to work out where agents fit in your SEO, book a free SEO strategy session.

Your iPhone does not choose the model

The iPhone app needs iOS 18 or later and connects to the same Bots, chats, routines and shared cloud computer as the desktop app, per the mobile docs.

The Bots run in the cloud, so a newer iPhone does not get a smarter Bot.

What the phone changes is how you work: short dictated prompts, quick approvals and results read on a small screen.

That is where the models you run beside it start to matter.

What a GoldieBench score means for a phone workflow

The methodology page says every model gets the same fixed prompt, one shot, and must return one self-contained HTML file, scored 0 to 10.

So a high score tells you a model can plan and ship a working build in a single pass.

It says nothing about screen control, approvals or the iPhone app itself.

Read the numbers as a guide to the models that write your briefs and review results, not as a rating of whatever Grok Bot routes to.

Grok on the bench, for context

The Grok model page lists a 256,000-token context and 5 gold, 1 silver and 1 bronze medal.

None of this tells you which model answers a given Grok Bot request, because the FAQ says routing can change per request.

It does tell you the Grok family is competitive on one-shot builds, which is a fair baseline.

Pages: the scores that matter when you read results on a phone

Grok Bot can build pages such as reports, trackers and dashboards, and on iPhone you open them from their card or the Bot Library (files and results docs).

GoldieBench has a Pages category, so it is the most relevant slice for this use.

On the Grok model page, Grok scores 9.0 with a gold medal on Landing Page and 9.0 with a gold on Webos Page, and 7.0 on Aipbpromo Page.

The home leaderboard lists Pages as the strongest category for GPT-5.6 Sol, GLM-5.2, Gemini 3.6 Flash and Claude Sonnet 5.

If a page from your Bot reads badly on the phone, ask for a narrower layout in the same chat rather than switching tools.

The planner: models that write the brief

Phone prompts tend to be short and dictated, which is exactly when a planner model earns its place.

Write the brief at the desk or in your Agent OS, then send it from the iPhone with the share sheet or Send Message shortcut.

On the home leaderboard, Claude Opus 5 sits at #2 with 8.27/10 across 50 tasks, and GPT-5.6 Sol at #4 with 8.16/10 across 50 tasks.

Hermes MoA, a panel of models merged by a chair, sits at #3 with 8.17/10 (models page).

Any of the three is a strong pick for turning a rough voice note into a tight task.

The Agent OS inside the AI Profit Boardroom is set up for exactly that planner step, before anything reaches the phone.

The checker: models that review before you tap Allow

Approval cards on the phone show the full request when you tap them, per the approvals docs.

For a long email draft or a page, paste it into a second model and ask what is wrong before you approve.

Use a different model from your planner, so the check is not marking its own homework.

GPT-5.6 Sol at 8.16/10 pairs well as a checker beside a Claude Opus 5 planner, or the other way round.

A fast model for quick drafts on the move

Not every phone task needs a flagship. Gemini 3.6 Flash scores 7.08/10 across 50 tasks at #18 on the home leaderboard, and its models-page card calls it faster and lighter on tokens.

Use a fast model for a one-line reply or a quick Custom Rule, and save the planner for briefs that drive real work.

Custom Rules on iPhone live in the Bot group of Settings, and each is a plain sentence with Allow automatically or Ask first.

A three-step iPhone workflow with these models

1. Plan: Claude Opus 5 or GPT-5.6 Sol turns your notes into a brief.
2. Send: share it into the Bot chat from the iPhone, with photos attached; HEIC shots show as images since v0.54.0 in the changelog.
3. Check: before tapping Allow on a big action, run the draft past your second model.

The Grok Bot changelog also notes that since v0.71.2, replies and questions notify your Mac or Windows PC instead of the phone while you use the desktop app.

So when you are at the desk, review there, and keep the phone for when you are away.

Background cards from routines expire after 30 minutes, so a checker step suits work you started yourself, where cards wait for your answer.

For routine output, have the Bot save a page you review later instead of asking for live approval.

What the bench cannot tell you about the iPhone app

• How reliable iPhone push alerts are: the docs say push delivery is rolling out gradually.
• How well a Bot uses your phone apps, such as Calendar, Reminders or Apple Home.
• Anything about billing: the app sells monthly individual plans through the App Store, and Apple handles them.

For those, the docs are the source, not a leaderboard.

If your plan is Premium, Super or Ultra, you already have Grok Bot, per Grok Bot on The X Subscription, so do not buy it again in the app.

Setup checklist

• Install from the App Store on iOS 18 or later.
• Sign in with the same Cursor account as your desktop.
• Pick one planner and one checker from the scores above.
• Add Ask first rules for anything public or costly.

For the desktop half of the setup, read the Grok Bot Mac model guide.

To get the Agent OS that wires your planner and checker together, join the AI Profit Boardroom.

Also On Our Network

• Agent Operatorsthe seven-step order for setting up the grok bot iphone app
• AI Income Deskwhat a small business owner can run from the grok bot iphone app
• AI Tool Verdictan eight-row scorecard rating the grok bot iphone app
• agentos.guidefive agent os prompts for running grok bot from an iphone
• aiprofitboardroom.comthe complete Grok Bot iOS guide on the AI Profit Boardroom blog

Pick a planner, pick a checker, and book a free SEO strategy session if you want my team to map the models around grok bot ios.