{"id":438,"date":"2026-09-22T08:10:39","date_gmt":"2026-09-22T00:10:39","guid":{"rendered":"https:\/\/aidashxp.com\/grok-4-7-review\/"},"modified":"2026-09-22T08:10:39","modified_gmt":"2026-09-22T00:10:39","slug":"grok-4-7-review","status":"publish","type":"post","link":"https:\/\/aidashxp.com\/en\/grok-4-7-review\/","title":{"rendered":"In-depth review of Grok 4.7: xAI\u2019s flagship reasoning model, starting at $2 for 500K context"},"content":{"rendered":"<p class=\"wp-block-paragraph\">September 21: xAI (SpaceXAI) officially launched <strong>Grok 4.7<\/strong>This is the next-generation flagship model following Grok 4.6 on August 12, officially positioned as \u201ca cutting-edge model for programming, agent tasks, and knowledge work.\u201d Its most striking feature is not raw capability\u2014but its pricing: $2 per million input tokens and $6 per million output tokens, unchanged from the previous generation yet only a fraction of GPT-5.6 Sol ($4\/$20) and <a href=\"https:\/\/claude.ai\" target=\"_blank\" rel=\"nofollow noopener\">Claude<\/a> Fable 5.1 ($10\/$50). In one sentence: xAI aims to directly capture the most lucrative market\u2014programming and knowledge work\u2014with a \u201ccutting-edge performance + rock-bottom pricing\u201d one-two punch.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Core capability: Specialized in \u201cmulti-hour\u201d hard tasks<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">The biggest change in Grok 4.7 lies not in parameter count (officially undisclosed; industry estimates place it at ~2.1 trillion parameters, roughly a 40% increase over Grok 4.6), but in<strong>training methodology<\/strong>xAI adopted a larger base model this time and conducted longer reinforcement learning (RL) training, with training data deliberately skewed toward problems requiring \u201chours to solve.\u201d<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">This yields three direct capability improvements:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Stronger self-verification<\/strong>The model excels at checking its own outputs, reducing \u201cplausible but incorrect\u201d hallucinations\u2014a critical advantage for extended programming tasks.<\/li>\n<li><strong>More stable long-context management<\/strong>With a 500K-token context window, maintains consistency across ultra-long documents and large codebases.<\/li>\n<li><strong>Native understanding of Grok Bot harness<\/strong>xAI specifically trained it to understand its own agent framework, resulting in better performance on conversational tasks and general knowledge work, as well as significantly improved document and PPT generation capabilities.<\/li>\n<\/ul>\n\n\n\n<h2 class=\"wp-block-heading\">Benchmark scores: Approaching the state of the art, but not universally dominant<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">xAI officially released a comparison table of Grok 4.7 versus Grok 4.6, GPT-5.6 Sol,<a href=\"https:\/\/claude.ai\" target=\"_blank\" rel=\"nofollow noopener\">Claude<\/a> and Fable 5.1. The conclusion is \u201capproaching the state of the art, with wins and losses across categories,\u201d not one-sided dominance:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Long-duration programming tasks<\/strong>46.3% (Grok 4.6: 40.4%, GPT-5.6 Sol: 41.7%, Fable 5.1: 51.8%) \u2014 a substantial improvement over the previous generation, yet still trailing Fable 5.1.<\/li>\n<li><strong>Terminal-Bench 4.0<\/strong>(Multi-hour agent terminal operations): 38.0%, nearly double Grok 4.6\u2019s 20.3%, representing the strongest generational improvement.<\/li>\n<li><strong>Engineering evaluation DE<\/strong>71.0%, surpassing Grok 4.6 (65.2%) and approaching GPT-5.6 Sol (72.7%).<\/li>\n<li><strong>Harvey Legal Agent Benchmark<\/strong>19.6%, substantially outperforming Fable 5.1 (6.7%) and GPT-5.6 Sol (2.5%), its strongest individual metric.<\/li>\n<li><strong>Clinical reasoning HealthBench<\/strong>56.7%, trailing Fable 5.1 (62.1%) and GPT-5.6 Sol (60.5%).<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">Additionally, xAI claims Grok 4.7 features<strong>a new safety guardrail stack<\/strong>scoring 62.4% on the LatchBio biosafety benchmark (leading), and permitting only 3.3% of hazardous dual-use prompts on its own HackerBench <a href=\"https:\/\/v0.dev\" target=\"_blank\" rel=\"nofollow noopener\">v0<\/a>.3, while rarely blocking legitimate safety research. Safety and compliance are emphasized more than ever compared to the previous generation.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">User experience and limitations<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Smooth onboarding path: Grok 4.7 is now integrated <strong><a href=\"https:\/\/cursor.com?referral=aidash\" target=\"_blank\" rel=\"nofollow noopener sponsored\">Cursor<\/a>Grok Build (x.ai\/build, free to try), Grok API<\/strong>and various third-party programming harnesses and model routers (e.g., OpenRouter). For speed, xAI also offers a \u201cfast variant\u201d that doubles output speed\u2014and doubles the price ($4\/$12).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Some reality checks are warranted:<strong>Chinese-language capability remains a weak spot<\/strong>Grok models have historically excelled at English and STEM tasks; Chinese generation quality lags noticeably behind domestic models (e.g., GLM,<a href=\"https:\/\/kimi.moonshot.cn\" target=\"_blank\" rel=\"nofollow noopener\">Kimi<\/a>,<a href=\"https:\/\/chat.deepseek.com\" target=\"_blank\" rel=\"nofollow noopener\">DeepSeek<\/a>). Second, it still falls short of first place on certain comprehensive benchmarks (e.g., long-horizon coding, clinical reasoning); if you prioritize \u201cabsolute strongest\u201d over \u201cbest value,\u201d Fable 5.1 or GPT-5.6 Sol remain superior.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Grok 4.7 vs. peer models\u2014head-to-head comparison<\/h2>\n\n\n\n<figure class=\"wp-block-table\"><table>\n<thead><tr><th>Model<\/th><th>context<\/th><th>Pricing ($\/M Token)<\/th><th>Core Positioning<\/th><th>This Site\u2019s Rating<\/th><\/tr><\/thead>\n<tbody>\n<tr><td><strong>Grok 4.7<\/strong><\/td><td>500K<\/td><td>2 \/ 6<\/td><td>Coding + knowledge work<\/td><td>8.7<\/td><\/tr>\n<tr><td><a href=\"https:\/\/aidashxp.com\/en\/claude-fable-5-1-review\/\">Claude Fable 5.1<\/a><\/td><td>\u2014<\/td><td>10 \/ 50<\/td><td>Enterprise agents<\/td><td>8.9<\/td><\/tr>\n<tr><td><a href=\"https:\/\/aidashxp.com\/en\/gemini-3-8-live-review\/\">Gemini 3.8 Live<\/a><\/td><td>\u2014<\/td><td>\u2014<\/td><td>Real-time voice<\/td><td>8.5<\/td><\/tr>\n<tr><td><a href=\"https:\/\/aidashxp.com\/en\/glm-5-3-flashx-review\/\">GLM 5.3 FlashX<\/a><\/td><td>\u2014<\/td><td>\u2014<\/td><td>High-Speed Multimodal<\/td><td>8.6<\/td><\/tr>\n<tr><td><a href=\"https:\/\/aidashxp.com\/en\/kimi-k3-amazon-bedrock\/\">Kimi K3<\/a><\/td><td>\u2014<\/td><td>\u2014<\/td><td>Domestic Inference<\/td><td>8.5<\/td><\/tr>\n<\/tbody><\/table><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">Plotting it on a coordinate system, Grok 4.7 occupies a shrewd position: capabilities approach Tier-1 levels, yet pricing is only a fraction of competitors\u2019. For developers and knowledge workers who need affordability, robustness, and sustained task performance, it may be one of the most balanced value propositions available today.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Overall Score<\/h2>\n\n\n\n<figure class=\"wp-block-table\"><table>\n<thead><tr><th>\u7ef4\u5ea6<\/th><th>Score<\/th><th>evaluate<\/th><\/tr><\/thead>\n<tbody>\n<tr><td>functional completeness<\/td><td>8.8 \/ 10<\/td><td>All-rounder for coding, agents, and knowledge work\u2014multimodal support remains weak<\/td><\/tr>\n<tr><td>\u6613\u7528\u6027<\/td><td>8.6 \/ 10<\/td><td><a href=\"https:\/\/cursor.com?referral=aidash\" target=\"_blank\" rel=\"nofollow noopener sponsored\">Cursor<\/a>Multiple access points: Grok Build, API\u2014fast onboarding<\/td><\/tr>\n<tr><td>Cost-effectiveness<\/td><td>9.0 \/ 10<\/td><td>$2\/$6 pricing is significantly lower than peers<\/td><\/tr>\n<tr><td>\u4e2d\u6587\u652f\u6301<\/td><td>7.8 \/ 10<\/td><td>Strong in English, weaker than domestic models in Chinese<\/td><\/tr>\n<tr><td>\u8f93\u51fa\u8d28\u91cf<\/td><td>8.7 \/ 10<\/td><td>State-of-the-art overall, but not top-ranked on all benchmarks<\/td><\/tr>\n<\/tbody><\/table><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Overall rating: 8.7\/10<\/strong><\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Frequently Asked Questions (FAQ)<\/h2>\n\n\n\n<h3 class=\"wp-block-heading\">Is Grok 4.7 free?<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">You can try Grok 4.7 for free via Grok Build (x.ai\/build); <a href=\"https:\/\/cursor.com?referral=aidash\" target=\"_blank\" rel=\"nofollow noopener sponsored\">Cursor<\/a>using Grok API or OpenRouter requires payment\u2014API pricing is $2 per million input tokens and $6 per million output tokens.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">What is the context window size of Grok 4.7?<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">Officially labeled as a 500K-token context window, sufficient to process ultra-long documents or large codebases in a single pass.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Grok 4.7 vs. GPT-5.6 Sol and<a href=\"https:\/\/claude.ai\" target=\"_blank\" rel=\"nofollow noopener\">Claude<\/a> Fable 5.1\u2014which is better?<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">Each has strengths: Grok 4.7 leads on Terminal-Bench and legal agent tasks and is the most affordable; GPT-5.6 Sol and Fable 5.1 excel at comprehensive tasks like long-horizon programming and clinical reasoning. Choose Grok 4.7 for value-for-money; choose the latter two for absolute top-tier performance.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Does Grok 4.7 support Chinese? How well does it perform?<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">It supports Chinese, but its Chinese generation quality lags behind domestic models such as GLM<a href=\"https:\/\/kimi.moonshot.cn\" target=\"_blank\" rel=\"nofollow noopener\">Kimi<\/a>,<a href=\"https:\/\/chat.deepseek.com\" target=\"_blank\" rel=\"nofollow noopener\">DeepSeek<\/a> \u2014its strengths lie in English and STEM tasks.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Want to Discover More Useful AI Tools? Explore Our <a href=\"https:\/\/aidashxp.com\/en\/ai-models\/\">AI Model Library<\/a> and <a href=\"https:\/\/aidashxp.com\/en\/compare-tools\/\">Tool Comparison Engine<\/a>or continue reading:<a href=\"https:\/\/aidashxp.com\/en\/claude-fable-5-1-review\/\">Claude Fable 5.1 Evaluation<\/a> \u00b7 <a href=\"https:\/\/aidashxp.com\/en\/glm-5-3-flashx-review\/\">GLM 5.3 FlashX benchmarking<\/a> \u00b7 <a href=\"https:\/\/aidashxp.com\/en\/reviews\/\">All Reviews<\/a>.<\/p>","protected":false},"excerpt":{"rendered":"<p>9 \u6708 21 \u65e5\uff0cxAI\uff08SpaceXA [&hellip;]<\/p>\n","protected":false},"author":0,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","ast-disable-related-posts":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"default","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":""},"categories":[6],"tags":[],"class_list":["post-438","post","type-post","status-publish","format-standard","hentry","category-ai-coding"],"_links":{"self":[{"href":"https:\/\/aidashxp.com\/en\/wp-json\/wp\/v2\/posts\/438","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/aidashxp.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/aidashxp.com\/en\/wp-json\/wp\/v2\/types\/post"}],"replies":[{"embeddable":true,"href":"https:\/\/aidashxp.com\/en\/wp-json\/wp\/v2\/comments?post=438"}],"version-history":[{"count":0,"href":"https:\/\/aidashxp.com\/en\/wp-json\/wp\/v2\/posts\/438\/revisions"}],"wp:attachment":[{"href":"https:\/\/aidashxp.com\/en\/wp-json\/wp\/v2\/media?parent=438"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/aidashxp.com\/en\/wp-json\/wp\/v2\/categories?post=438"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/aidashxp.com\/en\/wp-json\/wp\/v2\/tags?post=438"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}