Models·SpaceXAI

Grok 4.7: What It Does and What It Costs

x-ai/grok-4.7

Grok 4.7 is SpaceXAI’s model for coding, agent tasks, and knowledge work.

$2Input, per million tokens
$6Output, per million tokens
500KContext window
450KLongest answer

500K tokens of context — about 375,000 words, on a scale that ends at 2M.

What it can do

  • Reasoning
  • Vision
  • Tool use
  • Takes text
  • Takes image
  • Takes file
  • Gives text

Released 21 Sept 2026 · checked 3 Oct 2026.

About it

Grok 4.7 is SpaceXAI’s flagship model for coding, agent tasks, and knowledge work. It can work with text, images, and files, and supports reasoning and tools. Its context window is 500,000 tokens, with up to 450,000 tokens in one answer. Token prices are $2 per million tokens in and $6 per million out.

Review

In a line

Grok 4.7 is SpaceXAI’s flagship model for coding, agent tasks, and knowledge work. It succeeds Grok 4.6.

What it's good at

The model is aimed at coding and other knowledge work, including tasks where it uses tools to carry out a sequence of actions. It supports reasoning and is particularly suited to long-running software work and checking its own work.

It accepts text, images, and files, so it can handle more than written prompts alone.

Price in plain terms

Token use costs $2 per million tokens in and $6 per million tokens out. Tokens are small pieces of text the model reads or writes.

At those rates, reading a page of about 500 words costs about 0.13 cents. A chat turn with about 1,000 tokens in and 500 out costs about 0.5 cents; 1,000 such turns cost about $5.00.

Context and limits

Grok 4.7 can take in up to 500,000 tokens at once. It can produce up to 450,000 tokens in a single answer.

Its listed output is text.

Ways to use it

SpaceXAI is its maker. The model identifier is x-ai/grok-4.7.

Verdict

Grok 4.7 is a fit for people who need a model for coding, extended agent tasks, or knowledge work, especially when image or file input and tool use matter. Its large context window may suit work that needs substantial material considered at once. The listed token price is straightforward, and the rough examples put 1,000 typical chat turns at about $5.00. It may not suit someone seeking an output format other than text.

Verdict

Consider Grok 4.7 if your work involves coding, tool-based tasks, or knowledge work and you want to provide images or files as well as text. Its 500,000-token context window accommodates a large amount of input, while its per-million-token rates make usage costs estimable. Before choosing it, consider that its listed output is text only and that the price examples assume a particular input and output size; heavier use will cost more.

  • Supports reasoning and tool use.
  • Accepts text, images, and files.
  • 500,000-token context window and 450,000-token maximum output.
  • Strong fit for long-running software tasks and checking its work.
  • Its listed output format is text only.
  • Token use is billed by tokens read and written.
  • The example cost depends on using about 1,000 tokens in and 500 out per turn.
  • Pick it ifCoding and long-running software tasks; Tool-based work and knowledge tasks; Prompts that include images or files
  • Skip it ifYou need output in a format other than text.; You want a fixed per-turn price rather than token-based costs.

Pricing

Rates are $2 per million tokens in and $6 per million tokens out. The examples estimate about 0.13 cents to read a page, about 0.5 cents for a typical chat turn, and about $5.00 for 1,000 such turns.

Questions

What is Grok 4.7 for?

It is SpaceXAI’s flagship model for coding, agent tasks, and knowledge work.

Can Grok 4.7 work with images and files?

Yes. It accepts text, images, and files, and supports vision.

How much does Grok 4.7 cost?

Its rates are $2 per million tokens in and $6 per million tokens out.

What does a typical chat turn cost?

A turn with about 1,000 tokens in and 500 out costs about 0.5 cents. 1,000 such turns cost about $5.00.

How much can it read at once?

Its context window is 500,000 tokens.

How long can one answer be?

The maximum output is 450,000 tokens. Its listed output is text.

On these boards

More from SpaceXAI