The Once Times

AI

Grok 4 Unveiled: xAI’s Latest AI Breakthrough Shakes Up the Industry

Advanced Model Sets New Benchmarks in Performance and Accessibility

2 MIN READ

By Timmy

Published:

Updated:

xAI has officially launched Grok 4, the latest iteration of its cutting-edge artificial intelligence model, promising to redefine the landscape of AI technology. The release, highlighted during a keynote presentation, showcases remarkable improvements in speed, accuracy, and multimodal capabilities, positioning Grok 4 as a leader in the AI race.

The keynote revealed stunning performance metrics across various benchmarks. Grok 4 achieved a 100% score on the AIME25 test and an impressive 44.4% on the full set of Humanity’s Last Exam, outperforming competitors like Gemini 2.5 Pro and Claude Opus 4. In the ARC-AGI v1 Semi Private test, Grok 4 recorded a 66.6% accuracy, significantly ahead of its peers. These results show xAI’s commitment to pushing the boundaries of AI intelligence.

Humanity's Last Exam Benchmark of Grok 4
Humanity's Last Exam Benchmark of Grok 4
Benchmarks of Grok 4
Benchmarks of Grok 4
ARC-AGI result of Grok 4
ARC-AGI result of Grok 4

A standout feature is the introduction of SuperGrok Heavy, a premium upgrade available today for $300 per month. This tier offers access to Grok 4 Heavy, higher rate limits, and early access to new features, as outlined in xAI’s latest update. The company also highlighted enhancements in voice mode, promising quicker response times—up to 10x faster—making interactions more seamless for daily users.

For developer, the model introduces a 256k context window and real-time data search capabilities, enhancing its utility for enterprise-grade applications. Additionally, the integration of Frontier multimodal reasoning and advanced security features caters to a wide range of users, from individual innovators to large corporations.

xAI also has shared an exhilarating glimpse into its future plans following the successful launch of Grok 4. The journey begins with the ongoing rollout of Grok 4, which debuted this month and has already set new benchmarks in AI performance. Next, xAI is set to introduce a Coding Model in August 2025, designed to enhance programming capabilities for developers worldwide. This will be followed by a Multi-modal Agent in September 2025, expanding Grok’s ability to process and integrate various data types seamlessly. Rounding out the year, a Video Generation Model is slated for release in October 2025, poised to revolutionize content creation with AI-driven video technology.

xAI’s future roadmap
xAI’s future roadmap

You might also like

Editor's Picks