Text and chat · xAI

Grok

xAI's Grok models, in private mode, with vision, very long context and search on X. Grok 4.7 is the newest.

Private Moderate cost
Latest versionGrok 4.7September 21, 2026
Artificial Analysis Intelligence Index46 points#12 of 90 on Artificial Analysis Intelligence Index
Price$3.057 per million tokens written$0.00214 a page written
Key spec500K tokens of context

What sets it apart

Grok emphasizes long context and low output prices: Grok 4.7 takes a 500K window and text plus images for $2 / $6 per million tokens, well under the other frontier models. The earlier Grok 4.20 and 4.3 reach 1M tokens or more and are cheaper still.

  • $2 in, $6 out per million tokens
  • Reads text and images
  • Multi-agent variant (Grok 4.20)

Strengths and limits

Where it shines

  • Private mode despite being a closed model
  • Up to 2M tokens of context (Grok 4.20)
  • Can search posts on X as well as the web

Think twice when

  • Independent tests rate Grok 4.7 only slightly above 4.6
  • Grok 4.20 Multi-Agent cannot use Sub Rosa's agent tools
  • Grok 4.7: 500K token context
  • Vision, tool calling, X search

Use cases

Reading a large archive

Here are two years of customer emails. Find the five complaints that come up most and quote one example of each.

A 500K token window holds a very large pile of text in one go.

Describing a photo

This photo shows my garden wall with a crack. Describe the crack and say whether I should call a mason.

Grok 4.7 accepts images next to text.

Against its rivals

  • ClaudeClaude Sonnet 5.5 scores 56 against 46, at a higher output price.#1 of 90 on Artificial Analysis Intelligence Index (58 points)
  • GPTGPT-6 Sol costs more on output ($10 against $6) and scores 48, close to Grok 4.7 at 46.#4 of 90 on Artificial Analysis Intelligence Index (53 points)
  • GeminiGemini 3.8 Flash is faster and cheaper but scores 41.#5 of 90 on Artificial Analysis Intelligence Index (53 points)
Compare side by side ↗

Version history

Each version, dated, with what it changed over the one before.

Artificial Analysis Intelligence Index, version after versionHigher is better · Artificial Analysis

Hollow points are earlier versions no longer offered in Sub Rosa.

See the numbers
VersionReleasedpoints
Grok 4.32026-0425
Grok 4.202026-0426
Grok 4.52026-07-0839
Grok 4.62026-0844
Grok 4.72026-09-2146
  1. Grok 4.20
    • Reasoning model with text and image input at $1.25 / $2.50 per million tokens, plus a multi-agent variant.
    Source : artificialanalysis.ai
  2. Grok 4.3
    • 1M token context at the same $1.25 / $2.50 price, scoring 25 on the index.
    Source : artificialanalysis.ai
  3. Grok 4.5
    • Moves to a 500K window and $2 / $6 pricing, scoring 39 on the index against 25 for Grok 4.3.
    Source : artificialanalysis.ai
  4. Grok 4.6
    • Index score rises to 44 at the same price and window.
    Source : artificialanalysis.ai
  5. Grok 4.7
    • Index score of 46, described by xAI as its most capable model, with a May 2026 knowledge cutoff.
    Source : artificialanalysis.ai
  6. Grok Build 0.1Latest
    • Experimental compact model with a 256K window at $2 per million input tokens, per xAI’s model list. Release date not published.
    Source : docs.x.ai

Benchmarks

How to read these scores: what a benchmark tells you.

Artificial Analysis Intelligence Index: #12 of 90 models measured in the catalog. One score averaging about ten independent tests of reasoning, knowledge, math, coding and agent work, all run the same way for every model. Higher is better. Each model is listed at its highest reasoning effort. Two or three points apart is a close call; ten points is a different class of model.

Artificial Analysis Intelligence IndexHigher is better · Artificial Analysis, October 4, 2026
  1. 1. Claude Opus 5.5Claude58
  2. 2. Claude Sonnet 5.5Claude56
  3. 3. Claude Fable 5.1Claude53
  4. 4. GPT-6 AstraGPT53
  5. 5. Gemini 4 ArgonGemini53
  6. 6. GPT-6.1 SolGPT52
  7. 7. Muse WonderMuse52
  8. 8. Claude Opus 5Claude51
  9. 9. Claude Fable 5Claude50
  10. 10. GPT-6 SolGPT48
  11. 12. Grok 4.7Grok46
  12. 16. Grok 4.6Grok44
  13. 27. Grok 4.5Grok39
See the numbers
Modelpoints
1. Claude Opus 5.558
2. Claude Sonnet 5.556
3. Claude Fable 5.153
4. GPT-6 Astra53
5. Gemini 4 Argon53
6. GPT-6.1 Sol52
7. Muse Wonder52
8. Claude Opus 551
9. Claude Fable 550
10. GPT-6 Sol48
12. Grok 4.746
16. Grok 4.644
27. Grok 4.539

All 7 models in this family

ModelUsePrivacyPrice
Grok 4.7grok-4-7TextPrivate$3.057 per million tokens written$0.00214 a page written
Grok 4.6grok-4-6TextPrivate$3.057 per million tokens written$0.00214 a page written
Grok 4.5grok-4-5TextPrivate$3.057 per million tokens written$0.00214 a page written
Grok 4.3grok-4-3TextPrivate$1.272 per million tokens written$0.000891 a page written
Grok 4.20grok-4-20TextPrivate$1.272 per million tokens written$0.000891 a page written
Grok 4.20 Multi-Agentgrok-4-20-multi-agentTextPrivate$1.272 per million tokens written$0.000891 a page written
Grok Build 0.1grok-build-0-1TextPrivate$0.899 per million tokens written$0.000629 a page written

Specifications

SpecificationGrok 4.7Grok 4.6Grok 4.3
ReleasedSeptember 21, 2026August 2026April 2026
WeightsClosedClosedClosed
Acceptstext, imagestext, imagestext, images
Producestexttexttext
Context500K tokens500K tokens1,000K tokens
AbilitiesReads images, Uses Sub Rosa’s tools, Thinks before answering, Tuned for codeReads images, Uses Sub Rosa’s tools, Thinks before answering, Tuned for codeReads images, Uses Sub Rosa’s tools, Thinks before answering
Price$3.057 per million tokens written$3.057 per million tokens written$1.272 per million tokens written
PrivacyPrivatePrivatePrivate

Prices, context and lengths from the live catalog on October 16, 2026; parameters and licenses from the maker.