SpaceXAI Releases Grok 4.5: First Model Since Cursor Acquisition, Musk Calls It 'Opus-Class'

📰 Event Overview

On July 9, 2026, Elon Musk's SpaceXAI (formerly xAI) officially released its latest AI model, Grok 4.5. This release carries multiple special significances: it's not only SpaceXAI's latest effort in the intense AI race, but also the first model launched since the company completed its acquisition of AI code editor Cursor. Musk called Grok 4.5 an "Opus-class model" on social media, implying its performance has reached the industry's highest level.

This release occurred during the AI industry's unprecedented "72-hour whirlwind release week." Just one day before SpaceXAI released Grok 4.5, OpenAI had launched the GPT-5.6 series; on the same day as SpaceXAI's release, Anthropic also introduced its new "Reflect" feature. Three major AI labs密集 releasing new products in the same week highlights the white-hot intensity of current AI competition.

Even more noteworthy is the shift in Musk's attitude toward competitors. He publicly stated he was "clearly wrong about Anthropic" and called Anthropic the "clear AI leader." This statement sharply contrasts with his previous fierce rivalry with OpenAI CEO Sam Altman, also hinting at a possible forming partnership between SpaceXAI and Anthropic.

🔧 Grok 4.5 Technical Features

According to Reuters, Grok 4.5 is priced at $1.25/million input tokens and $2.50/million output tokens, placing it at a moderate-to-low price level among major frontier models. The model features configurable reasoning capabilities, allowing users to adjust reasoning depth based on task complexity, flexibly trading off between speed and intelligence.

Grok 4.5 excels in the following areas: instruction following, agentic tool use, and accuracy. These characteristics make it particularly well-suited for research, enterprise workflows, and general-purpose applications. Notably, "configurable reasoning" is an increasingly popular feature that allows users to reduce reasoning steps when fast responses are needed and increase reasoning investment when deep analysis is required.

As the first model since the Cursor acquisition, Grok 4.5 likely integrates Cursor's technical accumulation in code editing and developer tools. As one of the most popular AI code editors, Cursor has large amounts of high-quality data about developer workflows and code generation. Integration of these resources may have significantly improved Grok 4.5's performance on coding tasks.

🤝 SpaceXAI and Anthropic Relationship Shift

One of the most surprising developments this week is the significant improvement in the relationship between SpaceXAI and Anthropic. The two companies, once seen as rivals, now appear to be forming some kind of partnership. Musk publicly admitted he was "clearly wrong about Anthropic" and called it the "clear AI leader," a statement that attracted widespread attention in the tech world.

More substantive cooperation is reflected at the financial level: reportedly, SpaceXAI pays Anthropic $1.25 billion monthly for compute. This massive deal indicates that SpaceXAI depends on Anthropic's infrastructure to some extent for training and running its AI models. This relationship differs markedly from traditional competition models, reflecting the complex cooperation-competition dynamics in the AI industry.

The background for this relationship shift may be: training AI models requires massive compute, and even well-funded SpaceXAI needs to acquire sufficient computational resources through cooperation. As a leader in AI safety, Anthropic's technology and infrastructure hold high value. Through cooperation, SpaceXAI can quickly access advanced AI capabilities, while Anthropic gains a stable revenue source.

💼 Strategic Significance of Cursor Acquisition

SpaceXAI's acquisition of Cursor is key background for understanding the Grok 4.5 release. As one of the most popular AI code editors on the market, Cursor has millions of active users. Through acquiring Cursor, SpaceXAI gained:

• A massive developer user base and usage data
• Advanced code understanding and generation technology
• Mature IDE integration experience
• Deep insights into developer workflows
• Deep integration with the VS Code ecosystem

These resources are crucial for improving AI model coding capabilities. In current AI competition, coding ability has become one of the key metrics for measuring model performance. OpenAI specifically emphasized its coding benchmark scores when releasing GPT-5.6, and Anthropic's Fable 5 is also known for coding ability. By integrating Cursor's technology, SpaceXAI is positioned to quickly close the gap with competitors in the coding domain.

Additionally, Cursor's user base provides SpaceXAI with a direct channel for distributing AI models. In the future, Grok models may be directly integrated into the Cursor editor, providing developers with a seamless AI coding experience. This "model + tools" vertical integration strategy is similar to OpenAI integrating Codex into the ChatGPT desktop app.

📊 Market Competition Landscape Analysis

Grok 4.5's release further intensifies AI industry competition. According to price comparison data from Reuters, current major frontier model pricing is as follows:

From a pricing perspective, Grok 4.5 sits at a relatively low level with $1.25/$2.50 pricing, only higher than GPT-5.6 Luna. This pricing strategy indicates SpaceXAI is trying to attract users through cost-effectiveness, particularly in enterprise application scenarios. However, while Claude Fable 5 is the most expensive ($10/$50), it offers a 1-million-token ultra-long context window, providing unique advantages in scenarios requiring large document processing.

🤔 Frequently Asked Questions (FAQ)

📝 Summary

SpaceXAI's release of Grok 4.5 marks an important step for the company in the AI race. As the first model since the Cursor acquisition, Grok 4.5 likely integrates Cursor's technical advantages in code editing, giving it unique competitiveness in coding tasks. Musk positioning it as an "Opus-class" model shows SpaceXAI is determined to claim a place in the top-tier AI model arena.

Even more noteworthy is the shift in Musk's attitude toward Anthropic. From public competition to admitting the other party is the "clear AI leader," this transformation reflects the complex reality of coexisting cooperation and competition in the AI industry. The fact that SpaceXAI pays Anthropic $1.25 billion monthly for compute shows that even in the most competitive areas, business cooperation remains possible.

For users, Grok 4.5's release provides another high-quality option. Its relatively low price and configurable reasoning capability make it attractive in enterprise applications. As AI competition continues to intensify, users will benefit from more choices and better prices.