Grok Bot now runs on Anthropic's Claude Opus 5.5, replacing SpaceXAI/Cursor models for core intelligence

Grok Bot, SpaceXAI's consumer AI agent app, now uses Anthropic's Claude Opus 5.5 as its intelligence layer, replacing SpaceXAI/Cursor models that 9to5Mac describes as less capable. The change was reported on October 7, 2026.

What changed

According to 9to5Mac, Grok Bot's October update is a backend upgrade. Requests that were previously handled by SpaceXAI/Cursor models can now be handled by Claude Opus 5.5, which 9to5Mac describes as the newest version of one of Anthropic's most capable models. The report does not say whether Claude Opus 5.5 replaces the earlier models entirely or is offered alongside them. It also does not say whether users can choose between models.

The report quotes no statement from SpaceXAI or Anthropic. The announcement it references appears to be a post on X from a user named lauren (@poteto).

Background

• Launch: Grok Bot launched in August 2026 with dedicated iPhone and iPad apps. 9to5Mac reports that an iPad version and more affordable plans arrived in September.
• Platforms: The app is available for iPhone, iPad and Mac.
• Compute relationship: 9to5Mac reports that Anthropic became a SpaceX customer this year while shopping for compute to power Claude usage. The report presents this as context for the integration but does not describe the commercial terms.

Competitive context

Grok Bot is one of several consumer-focused AI agent apps. 9to5Mac names two rivals:

• Meta's Muse: a mostly free agent, now available on iPad.
• OpenAI's ChatGPT Dot: a proactive agent powered by Astra.

Not disclosed

The source does not provide any of the following:

• Claude Opus 5.5 pricing, either per 1M tokens or in Grok Bot subscription terms
• Context window size
• Benchmark scores comparing Claude Opus 5.5 with the models it replaces
• Whether Grok Bot's plan pricing changes as a result
• The terms of the Anthropic-SpaceX compute arrangement

What this means

A consumer agent app run by one AI company is now using a frontier model from another. That suggests SpaceXAI is prioritizing the product's quality over exclusive use of its own models, at least for this app. The reported compute relationship means Anthropic and SpaceX are now both customer and supplier to each other in some form. That is a more entangled arrangement than a standard API purchase, though the terms are unknown.

For users, the practical result depends on facts the report leaves out: how often Claude Opus 5.5 is used, whether higher-tier plans get more access, and how usage limits and costs change. Until SpaceXAI or Anthropic publish details, the claim that Grok Bot is "a lot smarter" rests on 9to5Mac's characterization rather than published evaluations.