llmrelay
// other

Grok 4.7

The latest xAI reasoning model. Released 21 September 2026.

Grok 4.7 is xAI’s latest reasoning model, released September 21, 2026. It runs on a new, larger base model than Grok 4.6, trained with a longer reinforcement-learning pass weighted toward tasks that take hours, and xAI’s own numbers show the biggest jump in the family so far: 46.3% on CursorBench 4.0 against Grok 4.6’s 40.4%, and 37.6% on Terminal-Bench 4.0 against 20.3%.

The commercial shape is unchanged from 4.6: $2/M input and $6/M output, the same 500k context window, and the same speed. xAI also serves a Fast variant (double the output speed at double the price) through Cursor and Grok Build, and 4.7 supports four effort levels — low, medium, high (default), and xhigh.

We sell it at half xAI’s list — $1.00 per million input tokens against $2.00, flat regardless of prompt length. xAI’s long-context billing doubles past 256k input tokens; ours does not, same flat-rate treatment as Grok 4.5 and 4.6.

Official list
llmrelay
Input / M tokens
$2.00
$1.00
Output / M tokens
$6.00
$3.00
Context window
—
500,000 tokens
Max output
—
500,000 tokens

What it costs you per month

Real-world budget scenarios. Numbers are simple sums — official list price vs llmrelay's 50% off tier.

Usage scenario
Official
llmrelay
Light coding (1M in / 200K out per month)
$3.20
$1.60save $1.60
Heavy Cursor / Cline user (50M in / 5M out per month)
$130.00
$65.00save $65.00
Production RAG (500M in / 20M out per month)
$1120.00
$560.00save $560.00

What it's good at

Best for

Pick Grok 4.7 when

  • +You want the latest xAI reasoning model. 4.7 is a new base model with a longer training run, not a refresh — the benchmarks moved more from 4.6 to 4.7 than they did from 4.5 to 4.6.
  • +Multi-hour coding and agent work. xAI positions 4.7 for long-running tasks: it works longer on hard problems and checks its own work more carefully than 4.6.
  • +Frontier reasoning at Sonnet money. At $1.00/$3.00 per million tokens it prices at Claude Sonnet 5 level (with cheaper output) while carrying frontier-model benchmarks and a 500k window.
  • +You want effort control. xhigh gives 4.7 more time on hard problems; low trades depth for latency.

Choose something else when

  • !You need more than 500k context. Claude Opus, Claude Sonnet, GPT-5.6 Sol and GPT-5.6 Terra all carry 1M windows, which is twice what Grok offers.
  • !Tool use and function calling are load-bearing. Grok supports it, but Claude’s and OpenAI’s implementations are more mature and more widely integrated.
  • !You want the longest-served id. 4.6 has been up longer; 4.7 is new. Both cost $1/$3 here, so pinning 4.6 is a stability choice, not a price one.
  • !Ordinary copy. Grok is a reasoning model. For hard Claude work use Opus 5 or Fable 5.1, not a writing SKU.

Questions people ask about Grok 4.7

How much does Grok 4.7 cost?

xAI’s list price is $2.00 per million input tokens and $6.00 per million output. Cursor’s on-demand pricing bills long-context requests at 2x past 256k input tokens, up to the 500k window. On llmrelay we charge $1.00 and $3.00 flat regardless of prompt length.

What changed from Grok 4.6 to 4.7?

A new, larger base model, a longer reinforcement-learning run on harder multi-hour tasks, better self-verification, and stronger long-context management. xAI also trained 4.7 to natively understand the Grok Bot harness. Price, speed, and the 500k context window are unchanged.

Should I use Grok 4.7 or Grok 4.6?

Both cost $1.00/$3.00 per million tokens here. Start with 4.7 — it is the same price and xAI’s benchmarks put it clearly ahead. Keep 4.6 only to pin an existing config.

What is the context window on Grok 4.7?

500,000 tokens input. xAI publishes no separate text output cap for 4.7 — output is bounded only by the context window — so we list the 500k ceiling as the output bound on this page.

Try Grok 4.7 at half the price

Free to create an account, no subscription, $10 minimum top-up — enough to run this model against a real task and judge quality yourself.

Get API key →

Compare Grok 4.7 against the alternatives

The comparisons this model appears in, the models nearest it on price, and the full rate card.