Perplexity's open-weight decision model that picks among given options for routing and checks in AI systems
Grok 4.7
New· 4Flagship Grok model for coding, agentic work and knowledge work, with a 500,000-token context window
About Grok 4.7
Released on 21 September 2026, Grok 4.7 is the newest flagship of the Grok family, built for coding, agentic work and knowledge work. Its developer documentation now credits SpaceXAI as the maker. The cost to run it is where the details become interesting, so most of this description is about what a buyer should check.
Context, input and reasoning settings
Grok 4.7 has a context window of 500,000 tokens, accepts text and images, and returns text, with no limit on text output stated. Its knowledge cutoff is May 2026. Four reasoning settings are offered: low, medium, high (the default) and xhigh. The company gives no parameter count, saying only that the model uses a new, larger base model.
Same rate card as Grok 4.6, different cost per task
The company's per-token rates are the same as those for Grok 4.6, with a higher rate for prompts over 200,000 tokens. That does not mean the same bill. Artificial Analysis, measuring independently, measured roughly 39.5 output tokens a second for Grok 4.7 where Grok 4.6 managed 60.4, and found Grok 4.7 used 240 million tokens to complete its index compared with 94 million for the earlier model. Its index score is 46 at the xhigh setting (version 4.3.2), two points ahead of Grok 4.6 measured the same way. The company says the model is served at the same speed as Grok 4.6, which the independent measurement does not bear out.
Variants and endpoints
A Fast variant exists too: the same model on quicker infrastructure at double the token rates. It is available only through Cursor and the maker's coding agent, and is absent from the public API. A US regional endpoint keeps inference in the United States and carries a surcharge of 10 percent.
A behavior to plan for
Through the Responses API, grok-4.7 returns its reasoning encrypted every time, whether or not a request asks for that. Anyone who logs or retains model output should account for that.
Vendor figures
The company's own results on CursorBench, DeepSWE and Terminal-Bench come from its own evaluation and are not independently verified.
TrustList summary
Grok 4.7 is the newest flagship model in the Grok line, released on 21 September 2026. It is aimed at coding, agentic tasks and knowledge work. The company's developer documentation names the model's maker as SpaceXAI, and the model takes text and images as input and returns text.
Main features
- Coding tasks: aimed at software work, including agentic tasks that run several steps.
- Knowledge work: aimed at document and research tasks alongside coding.
- Adjustable reasoning: four settings, low, medium, high (the default) and xhigh.
- Fast version: the same model on faster infrastructure, offered through Cursor and the company's own coding agent rather than the public API.
- US endpoint: a regional endpoint that keeps inference in the United States.
- Encrypted reasoning: on the Responses API, reasoning is always returned in encrypted form, even when not requested, which matters for teams that log model output.
Who it serves
Developers and teams working on coding, agentic and knowledge tasks, as the company positions it. No parameter count is given, though the company describes its base model as new and larger.
Key facts
- Developer: SpaceXAI, as named in the company's developer documentation
- Released: 21 September 2026
- Context window: 500,000 tokens
- Knowledge cutoff: May 2026
Summary written by TrustList from the business's own published information, Nov 2026.
Benchmarks & AI stats
| Benchmark | Official | Community avg |
|---|---|---|
| Artificial Analysis Intelligence Index v4.3.2llm | 46 index | - |
| Artificial Analysis Intelligence Index v5.0.1llm | 35 index | - |
“Official” values are editor-approved and feed the ranking. “Community avg” is the mean of member submissions (shown for transparency; it never affects the ranking until an editor approves a value).
Sign in to add a benchmark score for this model.
Grok 4.7 launches
Trust Score
An earned signal from verification, reviews, awards, transparency and engagement - the vendor can't buy it.
Updated 9/22/2026
What the business tells TrustList
- Founded
- 2026 (0 years) evidence sent
Request a demo or quote from Grok 4.7
Reviews
Questions & answers
No questions yet. Be the first to ask about Grok 4.7.
Grok 4.7 alternatives
Similar listings buyers compare against this one.
Aleph Alpha's open-weight English-German mixture-of-experts model for self-hosting, released under Apache 2.0 in October 2026
Cantina Security's first open-weight security model, a post-trained GLM-5.3-Flash for vulnerability research and exploit checks
JetBrains' open 12B mixture-of-experts coding model with 2.5B parameters active per token under Apache 2.0
Open-source 2B decision model from AWS's Strands Agents project that picks options and returns scores with a reliability figure
Anthropic's smallest and fastest Claude 5.5 model, available on its own API and three cloud platforms
Microsoft AI's text-to-speech model for keeping one voice across 23 languages, with voice cloning from short audio
Google's open multimodal embedding model that maps text, images, audio and video into one vector space, under Apache 2.0
More on TrustList
Everything here links back to the same verified catalogue. Pick your next stop.
- Write a reviewHelp other buyers decide
- CompaniesAgencies, consultancies and IT service providers, ranked by verified reviews.
- ProductsSoftware and SaaS with pricing, features, integrations and alternatives.
- ArticlesGuides, comparisons and research from the editorial desk and the community.
- AwardsAnnual recognition decided by verified reviews and an independent jury.
- LaunchesNew products and releases, voted up by the community every day.
- RequestsBuyers describe what they need; vendors respond directly.
- PeopleReviewers, authors and makers with public profiles.
- ComparePut up to four listings side by side before you shortlist.