Grok 4.5 Officially Announced: Features and Performance Compared to Competitor Flagship Models

Following the major transformation of rebranding to "SpaceXAI" that we introduced the other day, the company has officially announced its first new-generation flagship model, "Grok 4.5".

Stepping beyond the previous image of being an "AI strong on X (formerly Twitter)," its performance in autonomously handling business and development tasks has been significantly enhanced. In this article, we will briefly check the key points of what features Grok 4.5 has and how it differs from the latest models from other leading companies.

"Practical and Agentic Capabilities" evolved through joint training with Cursor

Grok 4.5 was trained in collaboration with "Cursor," the developer of a cutting-edge AI code editor.
As a result, the accuracy of its "Agentic AI" capabilities—where the AI itself autonomously understands the structure of large-scale source code and handles everything from bug discovery to correction and testing as a series of processes—has improved.
This capability is widely applied not only to programming but also to the automation of general office work (such as Grok Build), including complex Excel function processing and document creation.
The feature of this release is that it has gained the intelligence to actively support daily desk work while leveraging its traditional strength of being able to quickly access real-time information on X.

Differences from other companies' top-tier models (GPT-5.6 / Claude Opus 4.8)

Grok 4.5 has been developed with a strong awareness of top-class models from other companies, such as OpenAI's "GPT-5.6" and Anthropic's "Claude Opus 4.8."
While possessing advanced reasoning capabilities on par with them, it achieves both "overwhelming response speed" and "cost reduction." The pricing of $2 per 1 million input tokens and $6 for output boasts extremely high cost-performance compared to other companies' flagship models.
I have summarized the current account environment and the positioning differences compared to other companies' top models.

$$
\begin{array}{l|l|l}
\textbf{Comparison Item} & \textbf{OpenAI: GPT-5.6 (Limited Preview)} & \textbf{SpaceXAI: Grok 4.5 (Latest Announcement)} \\ \hline
\text{General Usage Environment} & \text{Limited availability to select companies due to government safety reviews} & \text{Widely deployed and available via X Premium and API} \\ \hdashline
\text{Development Strengths} & \text{Cybersecurity and robust system defense} & \text{Autonomous practical work and office automation via Cursor integration} \\ \hdashline
\text{Operating Costs} & \text{Relatively high data costs as a top-tier model} & \text{Low-cost route at $2 input / $6 output per 1 million tokens} \\ \hdashline
\text{Greatest Weapon} & \text{High reasoning performance} & \text{Real-time data from X + Overwhelming processing speed}
\end{array}
$$

Summary

Until now, Grok's position was that of a "slightly unique chatbot strong on the latest trends," but with the arrival of Grok 4.5, it has moved closer to the top-tier models of GPT and Claude.
While other companies' cutting-edge models are showing cautious deployment from the perspective of safety and governance, Grok 4.5, with its overwhelming processing speed and low cost, which can be practically called up at any time from the X app, may become a new option for information gathering and desk work in the future.

Grok is evolving from an "AI strong on the latest information" to an "AI that can be entrusted with practical work." With speed and cost-performance as its weapons, its presence in development and daily operations is likely to increase even further.

#TodaysAINews #Grok45 #SpaceXAI #GenerativeAI #FutureOfTechnology #DailyNotePost