The arena · AI model review
Grok 4.7
by xAI
xAI's most capable model, released September 2026: $2/$6 per 1M tokens, 500K context, text and image input
$6/1M out (<200K prompt), $12 above
Official xAI API price for grok-4.7: below 200K prompt tokens $2/1M input, $0.50/1M cached input, $6/1M output; requests whose prompt reaches 200K tokens are billed $4/$1/$12 for all tokens. Grok 4.7 Fast (Cursor and Grok Build only) costs 2x standard rates. Verified against docs.x.ai 2026-11-17.
xAI
500K tokens
$2/1M in (<200K prompt), $4 above
$6/1M out (<200K prompt), $12 above
text, vision (text output only)
No
What is Grok 4.7?
xAI's frontier model for coding, agentic tasks and knowledge work, released on the xAI API in September 2026 (September 21 per Artificial Analysis), after Grok 4.5 in July and Grok 4.6 in August. xAI calls it the most capable model it has built and recommends it for everything including code. 500K-token context, text and image input, text output, reasoning effort low, medium, high (default) or xhigh, knowledge cutoff May 2026. Price: $2/1M input, $0.50 cached, $6/1M output below 200K prompt tokens; from 200K, the whole request is billed at $4/$1/$12.
Grok 4.7 pros & cons
Pros
- xAI's current recommended model for all uses, coding included, per its own documentation
- Moderate price for a frontier-lab flagship: $2/1M input and $6/1M output, cached input at $0.50
- Text and image input with four reasoning effort levels, from low (about 11 s to the first token) to xhigh
- 500K-token context, enough for large codebases or document sets
- Also served on a US regional endpoint, and a faster Grok 4.7 Fast variant exists inside Cursor and Grok Build
Cons
- Below average on independent tests: 35 on the Artificial Analysis Intelligence Index v5, ranked 46th of 81, which calls it below average in intelligence but reasonably priced
- Slow and verbose at xhigh effort: about 50 s to the first answer token, 70 tokens/s, 148M output tokens and $7.77 per index task per Artificial Analysis
- Pricier than Grok 4.3 ($1.25/$2.50) with half its context (500K vs 1M), and requests from 200K prompt tokens are billed at double rates
- Grok 4.7 Fast is not on the public API: only in Cursor and Grok Build, at twice the standard rates
- No open weights and a single API provider
The arena’s verdict
Grok 4.7 is xAI's model to use today if you are on its API: it replaces Grok 4.3 as the house recommendation, coding included. The price stays moderate at $2/$6, but independent results are modest: Artificial Analysis ranks it 46th of 81 with 35 points, close to models that cost less, and at xhigh effort it is slow and very verbose. Coming from Grok 4.3, note that you pay more per token and lose half the context window (500K instead of 1M). For a comparable budget, test it against Kimi K3.1 Pro or Gemini 3.8 Flash before committing.
Thumbs up or thumbs down
Cast your verdict
Would you recommend Grok 4.7, or warn the crowd away?
Top Grok 4.7 alternatives
All alternativesGoogle workhorse September 2026: 1M context, $0.75/$3.75 promo pricing, 305 tokens/s output speed
Google's Nov 2025 frontier model: 1M context, first past 1500 Elo on LMArena, replaced by Gemini 3.1 Pro.
Meta's open-weight MoE duo: 17B active params, native image input, 10M-token context on paper
Compare Grok 4.7 head-to-head
Grok 4.7: frequently asked questions
How much does Grok 4.7 cost per 1M tokens?
Grok 4.7 costs $2/1M in (<200K prompt), $4 above per 1M input tokens and $6/1M out (<200K prompt), $12 above per 1M output tokens. Official xAI API price for grok-4.7: below 200K prompt tokens $2/1M input, $0.50/1M cached input, $6/1M output; requests whose prompt reaches 200K tokens are billed $4/$1/$12 for all tokens. Grok 4.7 Fast (Cursor and Grok Build only) costs 2x standard rates. Verified against docs.x.ai 2026-11-17.