Text & ChatFreemium

Grok

SpaceXAI's general-purpose AI assistant — chat, coding, agents, image and voice — built on the Grok model family.

4.4

Not re-verified since July 18, 2026

About Grok

Grok is the AI assistant from xAI, now part of SpaceX. It spans conversational chat (grok.com and Grok on X), agentic coding through Grok Build and Cursor, image generation via Grok Imagine, and voice. Its flagship is Grok 4.7 (September 21, 2026), at $2/$6 per million tokens with a 500K-token context, the same price as the Grok 4.6 and Grok 4.5 it followed, and the family is positioned as an efficient, low-cost alternative to the frontier models from Anthropic, OpenAI and Google, tuned for coding, agentic tasks and knowledge work.

Reviews (2)

Grok 4.7 Review: Same Price, Twice the Cost per Task

4.0

Grok 4.7 keeps Grok 4.6’s $2/$6 list price and beats it on every benchmark SpaceXAI published. The bill is another matter. Artificial Analysis measured $3.74 per Intelligence Index task at xhigh, about twice Grok 4.6’s $1.86 at high, for a score of 46 against 44. Claude Opus 5.5 at high scores 54 for $1.82. In our own build tests Grok 4.7 finished last in both rounds it entered.

Pros:

+Same $2/$6 list price as Grok 4.6 — with cached input at $0.50
+Beats Grok 4.6 on every benchmark SpaceXAI published
+Terminal-Bench 4.0 nearly doubles from 20.3% to 38.0%

Cons:

-$3.74 per index task at xhigh — about twice Grok 4.6 for two more points
-GPT-6 Astra and Claude Opus 5.5 at high both score higher for less per task
-Trails Claude Fable 5.1 by nearly 20 points on Terminal-Bench 4.0

Read full review →

Grok 4.5 Review: Opus-Class Coding, Far Cheaper

4.4

SpaceXAI's Grok 4.5 isn't the smartest model on the board — Fable 5 and GPT-5.5 still edge it — but it delivers roughly Opus-class coding at ~4.2× fewer tokens and a third of the price, and leads on long agentic runs. The best value in frontier coding today.

Pros: