Grok 4.7 launches: access, 500K context, and pricing limits

Grok 4.7 is available through the xAI API, Cursor, and Grok Build. Check its inputs, long-context pricing, Fast variant restrictions, and research limits.

Quick answer

SpaceXAI announced Grok 4.7 on September 21 for coding and knowledge work, with access through the xAI API, Cursor, and Grok Build. It accepts text and images and returns text, with a 500,000-token context window. The practical questions are which interface provides it, whether search is enabled, and which pricing tier applies. This is a source-based explainer, not a hands-on comparison.

Download Chat AI Opens the official App Store or Google Play for your device.

What changes from Grok 4.6?

SpaceXAI says Grok 4.7 uses a larger base model and longer reinforcement learning focused on extended tasks. The announcement highlights improvements in checking its work and producing documents and presentations. Those are provider claims, not results we have independently reproduced.

Sources: SpaceXAI

Standard access and Fast access are different

The documentation names grok-4.7 as the API model and lists Cursor and Grok Build as access routes. It supports text and image input, text output, and four reasoning settings: low, medium, high, and xhigh. High is the default.

  • Grok 4.7 Fast is the same model on faster infrastructure, charged at twice the standard token rates.
  • Fast is limited to Cursor and Grok Build, excludes Grok Build's free tier, and is not offered on the public xAI API.
  • These routes do not establish availability in every Grok-branded mobile experience or third-party app.

Sources: SpaceXAI

The headline price does not cover every long request

The published global API rates are in US dollars per million tokens. Below 200,000 prompt tokens, Grok 4.7 costs $2 for input, $0.50 for cached input, and $6 for output. At the pricing table's 200,000-token long-context threshold, those rates become $4, $1, and $12 respectively. The long-context rates apply to all tokens in that request, not only the portion above the threshold.

Sources: SpaceXAI

A large context window is not live research

SpaceXAI lists a May 2026 knowledge cutoff and says current information requires search tools. Supplying a long document gives the model material to analyze; it does not automatically update facts beyond that document. For current research, check that the interface actually enables search and open the cited sources.

Sources: SpaceXAI

An editorial exercise: reconcile a revised event brief

In an interface that explicitly offers Grok 4.7, try the fictional notes below before trusting a longer planning task. This is a proposed exercise with a hand-written reference answer, not a live Grok result or a test performed in Chat AI.

Example prompt — select and copy

Compare these notes. Note A: the room has 32 seats, attendance is 28 people, and the budget is $240. Note B, written later: attendance is now 36 people, including two hosts; room capacity and budget have not changed. Give the current attendance, seat shortfall, and budget. Identify what is unknown. Do not invent a venue, a price per seat, or a confirmed booking.

Frequently asked questions

What readers usually ask

Does Grok 4.7 generate images or spoken audio directly?

Its documented model output is text. Image input is supported, but image or audio generation is a separate capability supplied by other models or tools; do not assume an integration includes it.

Can I select Grok 4.7 Fast through the public xAI API?

No. The launch documentation limits Fast to Cursor and Grok Build. It is not included in Grok Build's free tier.

Is Grok 4.7 verified in Chat AI?

No. Chat AI's published catalog does not verify this exact model. This article explains the provider release and does not promise Grok 4.7 access through Chat AI.

Evidence

Sources

  1. Introducing Grok 4.7SpaceXAI · Primary source
  2. Grok 4.7 capabilities and access documentationSpaceXAI · Primary source
  3. API pricing and long-context thresholdsSpaceXAI · Primary source
  4. Model catalog and current-information limitationsSpaceXAI · Primary source