Grok 4.7

xAI's model for coding, engineering and office work, with a 500K-⁠token context window

About Grok 4.7

Grok 4.7 is xAI's model for coding and knowledge work, with about 2.1 trillion parameters. It reads text, images and files, takes up to 500,000 tokens in one conversation and keeps working on hard tasks for hours, checking its own work before it answers

Grok 4.7 scores 64.0% on EEBench, an electrical engineering test, ahead of Claude Fable 5.1 at 56.4% and GPT-⁠5.6 Sol at 39.4%, and 19.6% on Harvey's Legal Agent Benchmark, three times Fable 5.1's 6.7%. In coding it reaches 46.3% on CursorBench 4.0 and 71.0% on DeepSWE v1.1, up from 40.4% and 65.2% for Grok 4.6

On Overchat AI, Grok 4.7 works in one chat with Claude, GPT-⁠6 and Gemini, so you can hand the same task to another model mid-conversation

Developer

xAI

Released

Sep 21, 2026

Context window

500K tokens

Max output

450,000 tokens

Parameters

2.1T

Input

Text, images and files

Output

Text

Reasoning

Yes, with adjustable effort

Intelligence Index

35, #14 of 54 models

License

Proprietary

Available in

Grok API, Cursor, Grok Build, Amazon Bedrock, OpenRouter, Overchat AI

Modalities
Intelligence Index?: 35
Context window?: 500K tokens
Released: Sep 21, 2026

Grok 4.7 vs other models

Grok 4.7 ranks #41 on the Artificial Analysis Intelligence Index out of 54 models

Artificial Analysis Intelligence Index35#14 of 54 modelsBetter than 24% of 54 models

65Claude Fable 5.2
60Gemini 4 Carbon
59Claude Opus 5.5
57GPT-6 Astra
55GPT-6.1 Sol
50Muse Wonder Full
46Kimi K3.1 Pro
44Muse Wonder
35Grok 4.7
33DeepSeek V4.1 Flash

#5of 63 modelsAA-Omniscience Non-Hallucination RateHow often the model says it doesn't know instead of making up an answer80.0%
#9of 166 modelsGDPval-AATasks from real jobs, graded against experts' work60.8%
#19of 63 modelsAA-Omniscience AccuracyKnowledge questions answered correctly42.7%
#20of 136 modelsSciCodePython code for scientific computing57.4%

Source: Artificial Analysis

Compare Grok 4.7

Grok 4.7 vs Grok 4.6

Grok 4.7 scores higher than Grok 4.6 in 6 of 6 benchmarks. The widest lead is in Terminal-Bench 4.0: 37.6% against 20.3%. Grok 4.7 writes 81 tokens per second, Grok 4.6 writes 290

Benchmarks won: Grok 4.7: 6; Grok 4.6: 0
Average score: Grok 4.7: 49.2%; Grok 4.6: 40.5%
Output speed, tokens/s: Grok 4.7: 81; Grok 4.6: 290
Time to first token, s (log scale): Grok 4.7: 8.9; Grok 4.6: 32.0

Grok 4.6 aheadGrok 4.7 ahead

Terminal-Bench 4.0+17.3
EEBench+11.0
HealthBench Professional+8.2
CursorBench 4.0+5.9
DeepSWE v1.1+5.8
Harvey's Legal Agent Benchmark+3.8

Grok 4.7 vs GPT-5.6 Sol

Grok 4.7 scores higher than GPT-⁠5.6 Sol in 4 of 6 benchmarks. The widest lead is in EEBench: 64.0% against 39.4%. GPT-⁠5.6 Sol is ahead in HealthBench Professional: 60.5% against 56.7%

Benchmarks won: Grok 4.7: 4; GPT-5.6 Sol: 2
Average score: Grok 4.7: 49.2%; GPT-5.6 Sol: 42.4%

GPT-5.6 Sol aheadGrok 4.7 ahead

EEBench+24.6
Harvey's Legal Agent Benchmark+17.1
CursorBench 4.0+4.6
Terminal-Bench 4.0+0.3
DeepSWE v1.1−1.7
HealthBench Professional−3.8

Grok 4.7 vs Claude Fable 5.1

Grok 4.7 scores higher than Claude Fable 5.1 in 3 of 6 benchmarks. The widest lead is in Harvey's Legal Agent Benchmark: 19.6% against 6.7%. Claude Fable 5.1 is ahead in Terminal-Bench 4.0: 57.9% against 37.6%

Benchmarks won: Grok 4.7: 3; Claude Fable 5.1: 3
Average score: Grok 4.7: 49.2%; Claude Fable 5.1: 50.8%

Claude Fable 5.1 aheadGrok 4.7 ahead

Harvey's Legal Agent Benchmark+12.9
EEBench+7.6
DeepSWE v1.1+1.0
HealthBench Professional−5.4
CursorBench 4.0−5.5
Terminal-Bench 4.0−20.3

Best model by use case

Average benchmark scores of Grok 4.7 and 3 other models by use case

Coding
?Average of 3 tests
1. 1Claude Fable 5.159.9%
2. 2Grok 4.751.6%
3. 3GPT-5.6 Sol50.6%
4. 4Grok 4.642.0%
Knowledge work
?Average of 3 tests
1. 1Grok 4.746.8%
2. 2Claude Fable 5.141.7%
3. 3Grok 4.639.1%
4. 4GPT-5.6 Sol34.1%

Grok 4.7 benchmarks

Compare Grok 4.7 with Grok 4.6, GPT-⁠5.6 Sol and Claude Fable 5.1 in 6 benchmarks measuring coding and knowledge work

Order

Coding

CursorBench 4.0?46.3%−5.5
DeepSWE v1.1?71.0%−1.7
Terminal-Bench 4.0?37.6%−20.3

Knowledge work

EEBench?64.0%+7.6
Harvey's Legal Agent Benchmark?19.6%+3.8
HealthBench Professional?56.7%−5.4

Sources: xAI, Grok 4.7 launch post (Sep 21, 2026)

Grok 4.7 speed

Output speed is how fast the model writes (tokens per second, higher is better). Time to first token is how long you wait before the answer starts (lower is better)

Output speed and time to first token of 35 models

?Percentile

Time to first token, s (log scale)

Output speed, tokens/s (log scale)

Output speed and time to first token of 35 models
Models	Output speed, tokens/s (log scale)	Time to first token, s (log scale)
GPT-6 Luna	143	2.4
GPT-6.1 Terra	75	4.3
GPT-6.1 Sol	58	2.9
GPT-6 Astra	48	3.2
Claude Opus 5.5	88	1.9
Claude Opus 5	54	2.6
Claude Opus 4.8	59	1.3
Claude Fable 5.2	84	2.3
Claude Sonnet 5.5	102	1
Claude 5 Sonnet	215	10.5
Claude Sonnet 4.6	220	10.8
Claude Haiku 5.5	167	2.1
Gemini 4 Carbon	169	29.5
Gemini 4 Argon	91	6
Gemini 3.8 Flash	203	7.5
Gemini 3 Pro	270	11.2
Grok 4.7	81	8.9
Grok 4.6	290	32
Qwen 3.8 Max	81	30.6
DeepSeek V4 Pro	144	1.1
DeepSeek V4.1 Flash	422	2.2
Mistral Large 4	226	8.6
Hy4	214	37.7
ERNIE 5.5 Flash	405	2.7
Muse Wonder	980	8.7
Muse Spark	1129	6.2
Kimi K3.1 Pro	340	7.4
Kimi K3	338	5.9
GLM 5.3	197	1.2
Flint	194	12.3
Llama 4	81	0.7
Claude 4.5 Haiku	162	0.9
MiniMax M3	169	4.5
MiniMax M3.1 Flash	656	7.2
Kimi K2.7 Code	455	3.6

Typing test

Every model writes the same answer. The test waits the model's time to first token, then prints the text at the model's output speed and shows the total time

Answer length

Grok 4.781 tokens/s8.9 s to first token
Grok 4.6290 tokens/s32.0 s to first token

Source: Overchat AI, 2026-11-11

See how Grok 4.7 answers

Examples of Grok 4.7 answers to 3 prompts: Pick a power circuit, Write a backup script and Outline a pitch deck. Each answer is the unedited output of the model

I want to power a 12 V LED strip that draws 2 A from a 24 V supply. Should I use a resistor, a linear regulator or a buck converter? Explain with the numbers.

Grok 4.7

Use a buck converter. A resistor or linear regulator will each waste 24 W as heat (50% efficiency); a buck converter wastes only a few watts.

The load

LED strip: 12 V × 2 A = 24 W.
Supply: 24 V. You must drop 12 V while supplying 2 A.

Resistor

Drop 12 V at 2 A → R = 12/2 = 6 Ω.
Dissipation: 12 V × 2 A = 24 W (or I²R = 24 W).
Supply delivers 24 V × 2 A = 48 W; efficiency = 50%.
A 24 W resistor is large, runs very hot, and gives poor regulation if the strip current changes with temperature or brightness.

Linear regulator (e.g. a 12 V LDO or 7812-style part rated for ≥2 A)

Same electrical result: it drops (24-12) V × 2 A = 24 W as heat.
Input power still 48 W, efficiency 50%.
You need a substantial heatsink; the regulator will be hot and the supply is loaded at 2 A continuously.

Buck converter

A typical synchronous buck is ~90% efficient (often 85–95% depending on the module).
Input power ≈ 24 W/0.9 ≈ 27 W.
Input current ≈ 27 W/24 V ≈ 1.1 A.
Waste heat ≈ 2–4 W instead of 24 W — small heatsinking or none, far less stress on the 24 V supply, and stable 12 V output.

Bottom line: At 2 A the resistor and linear regulator each throw away as much power as the LEDs use. A buck converter is the only practical choice.

Write a bash script that backs up a folder to a dated .tar.gz file, keeps only the last 7 backups and logs each run. Then show how to run it every night at 2 am with cron.

Grok 4.7

bash

#!/usr/bin/env bash
# backup-folder.sh — archive a folder to a dated .tar.gz, keep the last 7, log each run.
set -euo pipefail

# --- config ---
SOURCE="${1:-/path/to/folder}"          # folder to back up (no trailing slash needed)
DEST="${2:-/path/to/backups}"           # directory that holds the archives
KEEP=7
LOG="${DEST}/backup.log"

# --- prep ---
if [[ ! -d "$SOURCE" ]]; then
  echo "Source is not a directory: $SOURCE" >&2
  exit 1
fi
mkdir -p "$DEST"

STAMP="$(date +%Y-%m-%d_%H%M%S)"
NAME="$(basename "$SOURCE")"
ARCHIVE="${DEST}/${NAME}_${STAMP}.tar.gz"

log() {
  # timestamped line to the log and to stdout
  printf '%s %s\n' "$(date '+%Y-%m-%d %H:%M:%S')" "$*" | tee -a "$LOG"
}

log "START backup of ${SOURCE} -> ${ARCHIVE}"

if tar -czf "$ARCHIVE" -C "$(dirname "$SOURCE")" "$NAME"; then
  log "OK wrote $(du -h "$ARCHIVE" | awk '{print $1}') ${ARCHIVE}"
else
  log "FAIL tar exited $?"
  exit 1
fi

# Keep only the newest KEEP archives for this folder name; delete the rest.
# shellcheck disable=SC2012
mapfile -t OLD < <(ls -1t "${DEST}/${NAME}_"*.tar.gz 2>/dev/null | tail -n +"$((KEEP + 1))" || true)
if ((${#OLD[@]})); then
  for f in "${OLD[@]}"; do
    rm -f -- "$f"
    log "PRUNE removed ${f}"
  done
else
  log "PRUNE nothing to remove (keeping up to ${KEEP})"
fi

log "DONE"

Make it executable and try a run:

bash

chmod +x backup-folder.sh
./backup-folder.sh /path/to/folder /path/to/backups

Nightly at 2:00 am with cron (use absolute paths; cron’s environment is minimal):

bash

crontab -e

Add:

cron

0 2 * * * /path/to/backup-folder.sh /path/to/folder /path/to/backups

That runs every day at 02:00. Output is also appended to /path/to/backups/backup.log. If you want cron’s own mail/stderr captured as well, you can append >> /path/to/backups/cron.log 2>&1 to the crontab line.

Outline a 6-slide pitch deck for a mobile app that helps roommates split rent and bills: a title and three short bullet points for each slide.

Grok 4.7

Slide 1: Title

SplitRent: fair bills, fewer roommate fights
One app for rent, utilities, and shared expenses
Built for households that want clarity without spreadsheets

Slide 2: The problem

Rent and bills are split unevenly or late
Venmo threads and spreadsheets get messy fast
Disputes over who owes what strain relationships

Slide 3: The solution

Create a household and add roommates in minutes
Log rent, utilities, groceries, and one-off costs
See who owes what and settle up in one tap

Slide 4: How it works

Set rent shares by room, percentage, or custom split
Auto-divide recurring bills and track due dates
Send payment requests and mark balances paid

Slide 5: Why it wins

Fairness rules that handle unequal rooms and incomes
Shared ledger everyone can trust
Reminders that reduce awkward money conversations

Slide 6: Ask

Seeking seed funding to launch iOS and Android
Use of funds: product, payments, and early growth
Goal: 50,000 active households in 18 months

More prompts to try with Grok 4.7

What Grok 4.7 is best for

Coding in your codebase
Grok 4.7 fixes bugs and builds features in real codebases, and it runs in Cursor and Grok Build. It scores 71.0% on DeepSWE v1.1 and 46.3% on CursorBench 4.0
Electrical engineering
Grok 4.7 sizes circuits, picks parts and checks power budgets, with the numbers shown. It scores 64.0% on EEBench, ahead of Claude Fable 5.1 and GPT-⁠5.6 Sol
Legal and office work
Grok 4.7 drafts documents, slide decks and contract reviews. It scores 19.6% on Harvey's Legal Agent Benchmark, against 6.7% for Claude Fable 5.1 and 2.5% for GPT-⁠5.6 Sol
Long multi-step tasks
Grok 4.7 keeps working on a hard task for hours and checks its own work before it answers. It nearly doubles Grok 4.6's score on Terminal-Bench 4.0, from 20.3% to 37.6%

How to use Grok 4.7 on Overchat AI

Start in the chat at the top of this page or in the Overchat AI app on the web, iOS or Android

01
Pick Grok 4.7
It is already picked in the chat on this page. In the Overchat AI app, open the model list next to the message field and pick Grok 4.7
02
Attach your code or a file
Add code, a document, a spreadsheet or a screenshot, and write what you need: the result, the format and what must stay the same
03
Keep going in the same chat
Ask a follow-up, ask for tests or a shorter version, or switch to another model without losing the conversation

Why use Grok 4.7 on Overchat AI

One account for chat, images and video, on every device

Superchat picked over Grok 4.7: one request done by Claude Opus 5.5 (the plan) and Nano Banana 2.1 (the poster)

Superchat AI agent

Superchat Agent automatically routes your message to the model best suited for the task. It can generate images and videos, use the AI tools built into Overchat AI, and learn from your interactions to get better over time

Message field with a PDF, a spreadsheet and a dashboard screenshot attached, Grok 4.7 picked and a voice message being recorded

Chat with images and files

Upload PDFs, spreadsheets, code and screenshots and ask Grok 4.7 about them. You can also dictate a message instead of typing it

An image made with Nano Banana 2.1 and a surfing video made with Seedance 2.5

Go beyond text

Generate images with Nano Banana and GPT Image and videos with Seedance and Kling in the same account, with the same credits as chat

On the web and iPhone

Start a chat on your laptop and finish it on your phone. The Overchat AI app has the same models, and your chats stay in one account on every device

Grok models on Overchat AI

Compare 2 Grok models on Overchat AI by output speed, time to first token and the price of one message

Models

Output speed, tokens/s

First token, s

Credits per message?

Grok 4.7This page

8.9

Grok 4.6

290

32.0

Source: Overchat AI, 2026-11-11

Explore other models

Models marked Soon are coming to Overchat AI; the rest are available now

More models

Learn more about AI models

All articles

We tested Gemini 4 Argon against Claude Opus 5.5, GPT-⁠6 Astra and Muse Wonder

We tested Gemini 4 Argon against Claude Opus 5.5, GPT-⁠6 Astra and Muse Wonder on three coding tasks. See benchmarks, speed and how to use it

Muse Wonder vs Muse Spark: Meta's New Model Tested

We tested Muse Wonder against Muse Spark 1.3 on logic, writing and coding tasks. See benchmarks and which Meta model is better

10 Best All-in-One AI Platforms in 2026: Why One AI Subscription Beats Many

Compare the 10 best all-in-one AI platforms in 2026, from Overchat AI to Poe and Magai, with prices, free plans and the newest GPT and Claude models

Grok 4.7 FAQ

If you have any questions,
reach out to us at hello@overchat.ai

Updated November 11, 2026

Grok 4.7 is xAI's model for coding and knowledge work, released on September 21, 2026. It has about 2.1 trillion parameters, reads text, images and files and takes up to 500,000 tokens in one conversation

Grok 4.7 is strongest at engineering and legal work: it scores 64.0% on EEBench and 19.6% on Harvey's Legal Agent Benchmark, ahead of Claude Fable 5.1 and GPT-⁠5.6 Sol. Claude Fable 5.1 is ahead at long coding tasks in a terminal, with 57.9% on Terminal-Bench 4.0 against 37.6% for Grok 4.7

Yes. Grok 4.7 scores 71.0% on DeepSWE v1.1 and 46.3% on CursorBench 4.0, and it runs in Cursor and in Grok Build, xAI's coding agent. On Overchat AI you can paste your code or attach files and ask it to fix or extend them

Grok 4.7 is built on a larger base model and trained longer on tasks that take hours. It scores 37.6% on Terminal-Bench 4.0 against 20.3% for Grok 4.6, 46.3% on CursorBench 4.0 against 40.4% and 64.0% on EEBench against 53.0%

Yes. Grok 4.7 reads images, screenshots and files along with text, and writes text and code. To make pictures or videos on Overchat AI, switch to Grok Imagine in the same app

Grok 4.7 reads up to 500,000 tokens in one conversation, about 375,000 words, and writes up to 450,000 tokens in one answer

Type your message in the chat at the top of this page, attach a file if you need to and press Send. After you sign in, Grok 4.7 answers in the Overchat AI chat, where you can keep the conversation going on the web, iOS or Android and switch to another model at any time

More tools

AI Code Generator AI Text Summarizer AI Playground Claude Fable 5.2 GPT-6.1 Terra Gemini 4 Carbon Kimi K3.1 Pro Hy4

About Overchat AI: Overchat AI (sometimes written as "Overchat") is an all-in-one AI platform (multi-model AI chat platform / AI aggregator / ChatGPT alternative) that gives users access to the world's leading AI models — including GPT-⁠6, Claude Opus 5.5, Gemini 3.8, Grok 4.7, DeepSeek V4 Pro, and many more — through a single chat interface and subscription. Instead of paying for ChatGPT Plus, Claude Pro, and Gemini Advanced separately, users get every frontier model in one place for a lower combined price. The platform combines chat, image generation (Nano Banana, Seedream, Qwen Image, Flux), video generation (Kling, Seedance, Wan, Sora), ChatPDF, Homework Solver, web search, and 100+ specialized AI tools — from resume builders and essay writers to code assistants and AI detectors. Files, PDFs, images, and links can be attached to any conversation. Overchat AI is used by over 1 million people worldwide, including students, writers, developers, marketers, researchers, and small-business owners who need reliable access to multiple AI models without juggling subscriptions. It is available on the web at overchat.ai and on iOS

Grok 4.7

About Grok 4.7

Grok 4.7 vs other models

Compare Grok 4.7

Grok 4.7 vs Grok 4.6

Grok 4.7 vs GPT-5.6 Sol

Grok 4.7 vs Claude Fable 5.1

Best model by use case

Coding

Knowledge work

Grok 4.7 benchmarks

Coding

Knowledge work

Grok 4.7 speed

Output speed and time to first token of 35 models

Typing test

See how Grok 4.7 answers

The load

Resistor

Linear regulator (e.g. a 12 V LDO or 7812-style part rated for ≥2 A)

Buck converter

More prompts to try with Grok 4.7

What Grok 4.7 is best for

Coding in your codebase

Electrical engineering

Legal and office work

Long multi-step tasks

How to use Grok 4.7 on Overchat AI

Pick Grok 4.7

Attach your code or a file

Keep going in the same chat

Why use Grok 4.7 on Overchat AI

Superchat AI agent

Chat with images and files

Go beyond text

On the web and iPhone

Grok models on Overchat AI

Explore other models

Learn more about AI models

We tested Gemini 4 Argon against Claude Opus 5.5, GPT-⁠6 Astra and Muse Wonder

Muse Wonder vs Muse Spark: Meta's New Model Tested

10 Best All-in-One AI Platforms in 2026: Why One AI Subscription Beats Many

Ready to try Grok 4.7?

Grok 4.7 FAQ

What is Grok 4.7?

Is Grok 4.7 good?

Is Grok 4.7 good for coding?

What is the difference between Grok 4.7 and Grok 4.6?

Can Grok 4.7 read images?

What is the context window of Grok 4.7?

How do I use Grok 4.7 on Overchat AI?

More tools