惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
大猫的无限游戏
大猫的无限游戏
博客园 - 聂微东
Jina AI
Jina AI
The Cloudflare Blog
V
Visual Studio Blog
博客园_首页
量子位
酷 壳 – CoolShell
酷 壳 – CoolShell
博客园 - 【当耐特】
爱范儿
爱范儿
博客园 - 三生石上(FineUI控件)
小众软件
小众软件
博客园 - 司徒正美
阮一峰的网络日志
阮一峰的网络日志
Last Week in AI
Last Week in AI
V
V2EX
钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知
博客园 - 叶小钗
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
WordPress大学
WordPress大学
宝玉的分享
宝玉的分享
T
Tailwind CSS Blog
博客园 - Franky

Data Studios ‧Exafin

Claude Code With Opus 4.7: Code Quality, Agentic Editing, Validation Loops, and Workflow Reliability in Modern OpenRouter for Production Apps: Routing, Fallbacks, Uptime, and Provider Resilience Across Multi-Model AI Infr Claude Opus 4.7 for Coding: Agentic Development, Debugging Workflows, Code Validation, and Professional Limits in Autonomous Software Engineering ChatGPT 5.5 Pro: Pricing, Context Window, Reasoning Depth, and Professional Limits for Advanced AI, Finance, R Grok 4.20 vs Grok 4: Speed, Reasoning, Access, Pricing, and Model Differences for API and Product Workflows Claude Code Project Setup: CLAUDE.md, Memory Files, Rules, and Team Conventions for Reliable Repository Workfl OpenRouter for OpenAI-Compatible Apps: Migration, SDK Portability, and Provider Switching Across Multi-Model W Claude Opus 4.7 for Difficult Prompts: Instruction Following, Consistency, and Complex Reasoning Across High-C ChatGPT 5.5 for Scientific Work: Data Analysis, Research Reasoning, and Complex Problem Solving Across Multi-S Grok Structured Outputs: JSON, Function Calling, Tool Use, and Automation-Ready Responses for Production Applications Claude Code Quality Reports: Regressions, Caching Issues, and Reliability Lessons for Agentic Coding Tools OpenRouter Analytics: Usage Tracking, Budget Controls, and Multi-Model Cost Visibility Across AI Workflows Claude Opus 4.7 Pricing: API Costs, Plan Access, Context Limits, and Usage Trade-Offs for Long-Context Workflows ChatGPT 5.5 System Card: Safety, Limitations, Evaluations, and Enterprise Relevance for Agentic AI Workflows Grok 4.20 Context Window: Long Inputs, Files, Collections, and Retrieval Workflows Across 2M-Token Reasoning S Claude Code GitHub Actions: Automated Reviews, CI Workflows, and Repository Automation Across Event-Driven Dev OpenRouter Tool Calling: Function Schemas, Structured Responses, and App Integration Across Production AI Work Claude Opus 4.7 for Computer Use: Browser Actions, Tool Execution, and Task Automation Across Agentic Workflow ChatGPT 5.5 for Enterprise Work: Agents, Professional Analysis, and Document-Heavy Tasks Across Governed Business Workflows Grok Imagine API: Image Generation, Video Generation, and Creative Media Workflows Across Programmable Visual Production Claude Code Slash Commands: /compact, /review, Fast Mode, and Terminal Productivity Across Agentic Coding Work OpenRouter Model Discovery: Providers, Benchmarks, Context Windows, and Effective Pricing Across Multi-Model API Workflows Claude Opus 4.7 for Enterprise Teams: Task Reliability, Workflow Automation, and Codebase Support Across Agentic Development Systems ChatGPT 5.5 vs ChatGPT 5.4: Pricing, Tools, Context Window, and Performance Differences for API and ChatGPT Wo Grok 4.20 for Coding: Technical Prompts, Tool Calling, and Developer Workflows Across Agentic Software Systems Claude Code Permissions: Safe Command Execution, Project Control, and Developer Guardrails Across Agentic Codi OpenRouter Video Inputs: Multimodal Models, File Handling, and Practical API Workflows for Video Understanding Claude Opus 4.7 for Long-Context Work: Large Files, Repositories, and Multi-Document Projects Across 1M-Token ChatGPT 5.5 in Codex: Coding Agents, Debugging, and Software Development Workflows Across Repository Context a Grok Voice API: Real-Time Conversation, Transcription, and Voice Agent Workflows Across Speech-to-Speech Syste
Grok Pricing: Subscription Tiers, API Token Costs, and Mo...
Michele Stef · 2026-04-25 · via Data Studios ‧Exafin

Grok is sold through several different commercial layers, and the confusion around pricing usually begins when those layers are treated as if they were one product with one billing model.

In practice, Grok access is divided between consumer subscriptions on X, paid Grok plans on Grok.com, and usage-based API billing through xAI for developers and companies building with Grok models programmatically.

That means the right way to read Grok pricing is not as a single price list, but as a set of access products and infrastructure products that serve different users, different workflows, and different purchase decisions.

·····

X subscriptions provide Grok access as part of a broader platform bundle.

X’s public subscription documentation lists Basic at $3 per month or $32 per year, Premium at $8 per month or $84 per year, and Premium+ at $40 per month or $395 per year on web, while also noting that pricing can vary by region and platform.

Those prices describe X subscription tiers rather than a standalone developer pricing system for Grok, which is why they should be understood as access bundles that include a wider set of platform features beyond AI usage alone.

For Grok specifically, the strongest public tie between subscription tiers and advanced model access comes from xAI’s Grok 4 launch materials, which state that Grok 4 was available to Premium+ and SuperGrok subscribers.

X Help also states that Premium+ includes higher limits on Grok, reinforcing the idea that Premium+ is the main X-native subscription tier for stronger Grok access rather than a general social subscription with only incidental AI usage.

........

X Subscription Pricing and Grok Positioning

Plan

Public Web Price

Main Commercial Role

Grok Positioning

Basic

$3 per month or $32 per year

Entry subscription on X

Not positioned as the main advanced Grok tier

Premium

$8 per month or $84 per year

Mid-tier X subscription

Broader subscription access with platform features

Premium+

$40 per month or $395 per year

Highest public X subscription tier

Higher Grok limits and premium Grok access positioning

·····

Grok.com subscriptions separate direct Grok access from X-native subscription access.

xAI’s main product site says users can unlock a SuperGrok subscription on Grok.com, which establishes Grok.com as a separate commercial surface rather than just an extension of X subscription billing.

xAI also introduced SuperGrok Heavy as a higher-tier offering that provides access to Grok Heavy with materially higher rate limits, and in the Grok 4 launch materials the company tied SuperGrok Heavy specifically to Grok 4 Heavy.

That matters because it shows that Grok model access is not determined only by whether a user is paying for Grok in some general sense, but by which subscription surface and which specific tier they are using.

xAI’s management documentation reinforces this structure by listing SuperGrok and SuperGrok Heavy as distinct license types for managed environments, which suggests that the distinction is not only consumer-facing but also part of how Grok access is organized for team and business usage.

A key limitation in the public materials is that the sources reviewed here clearly show the existence and role of SuperGrok and SuperGrok Heavy, but they do not expose a reliable plain-text public dollar price for those plans in the same way X Help does for X Premium tiers.

........

Grok.com Subscription Logic and Model Access

Tier

Commercial Surface

Access Logic

Model Positioning

SuperGrok

Paid Grok subscription

Access path for higher-tier Grok usage including Grok 4

SuperGrok Heavy

Higher-tier paid subscription with higher limits

Access path for Grok 4 Heavy

Managed SuperGrok licenses

Team and business environments

Admin-managed enhanced access

Extends Grok subscription logic into organizational use

·····

xAI API pricing uses a separate token-based billing model for developers.

The xAI API is a distinct commercial product from X subscriptions and Grok.com subscriptions, and its pricing is based on metered model usage rather than monthly access bundles.

xAI’s public API pricing page shows Grok 4.20 non-reasoning at $2.00 per million input tokens, $2.00 per million cached input tokens, and $6.00 per million output tokens.

The same public pricing materials show Grok 4.1 Fast Reasoning at $0.20 per million input tokens, $0.20 per million cached input tokens, and $0.50 per million output tokens, which indicates that xAI is differentiating API prices not only by model family but also by performance profile and intended workload.

xAI’s developer documentation also makes clear that API usage is set up through an xAI account funded with credits, which confirms that API access is an infrastructure purchase path rather than a benefit automatically bundled into consumer Grok subscriptions.

The cost story becomes even more important when tools are involved, because xAI’s docs note that tool usage can introduce additional pricing components on top of token consumption.

........

Published xAI API Pricing for Selected Grok Models

Model

Input Tokens

Cached Input Tokens

Output Tokens

Grok 4.20 Non-Reasoning

$2.00 per million

$2.00 per million

$6.00 per million

Grok 4.1 Fast Reasoning

$0.20 per million

$0.20 per million

$0.50 per million

·····

Model access changes depending on whether Grok is used through X, Grok.com, or the xAI API.

One of the most important distinctions in Grok pricing is that model access follows the surface through which the user is buying access rather than following a single universal entitlement across the whole xAI ecosystem.

On consumer surfaces, xAI has publicly tied Grok 4 access to Premium+ and SuperGrok, while Grok 4 Heavy has been tied to SuperGrok Heavy.

On the API side, access is determined through xAI’s developer platform, model catalog, account credits, and rate-limit structure rather than through consumer subscriptions.

xAI’s business materials add another layer by describing organizational access to advanced Grok models, including Grok 3, Grok 4, and Grok 4 Heavy, through managed licenses and business offerings.

There is also a regional nuance on the API side, since xAI’s regional endpoint documentation says the global endpoint can access all models available to the team through automatic routing, while region-specific endpoints may have different model availability and may fail rather than reroute if a region cannot serve the request.

........

How Model Access Differs by Purchase Path

Access Path

Billing Logic

Main User Type

Model Access Logic

X subscriptions

Monthly or annual subscription

Consumers inside X

Grok access tied to X tier and usage limits

Grok.com subscriptions

Paid direct Grok plans

Consumers and power users using Grok directly

Higher Grok tiers unlock stronger model access and higher limits

xAI API

Token-based metered billing

Developers and companies

Model access tied to developer account, credits, and API configuration

xAI business licensing

Managed commercial access

Teams and enterprises

Advanced model access tied to organizational licensing and admin controls

·····

Grok pricing only makes sense when access products and infrastructure products are treated separately.

The most accurate way to read Grok pricing is to separate subscription access from developer infrastructure, because X Premium+ and Grok.com plans are selling user-facing access, quotas, and higher-tier product privileges, while the xAI API is selling metered model usage for software integration and production workloads.

That distinction also explains why there is no single answer to the question of what Grok costs, since the answer changes depending on whether the buyer wants interactive access inside X, direct access through Grok.com, or programmatic access through the xAI API.

It also explains why model access differs across products, because Premium+, SuperGrok, SuperGrok Heavy, and API accounts are not interchangeable commercial states even when they all relate to Grok.

For end users, the central question is which subscription surface provides the right level of Grok access and usage capacity.

For developers, the central question is which Grok model and token price make sense for the intended workload, especially when output volume, reasoning mode, tool usage, and endpoint availability all affect the real cost of deployment.

That is the real structure behind Grok pricing and model access.

·····

FOLLOW US FOR MORE.

·····

DATA STUDIOS

·····

·····