Grok 4.7

// Overview

Grok 4.7 is SpaceXAI's Grok flagship, announced on September 21, 2026 and described by the company as its most powerful model for coding and knowledge work. SpaceXAI says it runs on a larger base model than Grok 4.6, with extended reinforcement learning on harder and longer-running tasks, and that the result works longer on difficult problems, verifies its own output more reliably, and manages extended context better. The company also gave it native understanding of the Grok Bot interface.

The model keeps the 500,000-token context window of Grok 4.6, accepts text and image input, returns text, and has a knowledge cutoff of May 2026. Its API model id is grok-4.7. Reasoning effort is configurable as low, medium, high (the default), or xhigh, and the model supports function calling, web search, X search, and code execution through both the Responses API and Chat Completions.

Pricing did not move with the upgrade: $2.00 per 1M input tokens and $6.00 per 1M output tokens for prompts under 200K tokens, the same rates SpaceXAI charged for Grok 4.6, with a fast variant served at twice the speed for twice the cost. Grok 4.7 is available in Cursor, Grok Build, the SpaceXAI API, third-party coding harnesses, model routers and cloud platforms, and GitHub added it to Copilot on launch day for Pro, Pro+, Max, Business, and Enterprise plans.

// Benchmarks

This model's scores

1. DeepSWE v1.171%
2. CursorBench 4.046.3%
3. EEBench64%
4. Terminal-Bench 4.038%
5. HealthBench Professional56.7%
6. Harvey Legal Agent Benchmark19.6%
7. AA Intelligence Index46

Scores on a 0–100 scale (25-point gridlines); higher is better. Each benchmark links to its published source.

// Pricing

Rates apply to prompts under 200K tokens; prompts of 200K tokens or more are billed at $4.00 input, $1.00 cached input, and $12.00 output per 1M tokens. A fast variant is served at twice the speed for twice the cost.

Pricing source ↗

// Strengths

• Beats Grok 4.6 on every benchmark SpaceXAI published — CursorBench 4.0 46.3% vs 40.4%, DeepSWE v1.1 71.0% vs 65.2%, Terminal-Bench 4.0 38.0% vs 20.3%
• Leads GPT-5.6 Sol and Fable 5.1 on EEBench electrical engineering at 64.0%
• Strongest published score on the Harvey legal agent benchmark in its comparison set at 19.6%
• 500,000-token context window with text and image input, unchanged from Grok 4.6
• No price increase over Grok 4.6: $2.00 / $6.00 per 1M input/output tokens under 200K-token prompts

// Best for

• Reach for it for coding agents in Cursor, Grok Build or GitHub Copilot that must hold a task across many steps.
• Reach for it for engineering-heavy knowledge work — it posts the highest published EEBench score in its comparison set.
• Reach for it for legal and document-heavy agent runs, where SpaceXAI reports a large lead on the Harvey benchmark.
• Reach for it when you are already on Grok 4.6 and want the gains without changing your per-token budget.

// How to access

// Grok (flagship) — every version

The full lineage of the Grok (flagship) line, newest first. Every version has its own page — click any to compare specs, benchmarks and pricing.