xAI Releases Grok 4.5 Optimized for Coding and Agentic Tasks

xAI has released Grok 4.5, a new frontier model specifically designed for coding and agentic tasks. Developed in partnership with the coding platform Cursor, the model is positioned as a high-performance alternative to industry leaders like Anthropic’s Opus and OpenAI’s GPT series. Rather than competing solely on absolute benchmark supremacy, xAI is emphasizing its efficiency, speed, and cost-effectiveness, with pricing set at $2 per million input tokens and $6 per million output tokens.

Technical evaluations from Artificial Analysis place Grok 4.5 among the top-tier models, ranking it fourth on both general intelligence and agent-specific benchmarks. The model features 1.5 trillion parameters and currently operates with a 500k context window, though xAI plans to increase this to 1 million tokens soon. Early feedback from developers highlights its speed and decisiveness in coding workflows, though some users noted potential performance degradation during very long sessions.

The release is significant because it signals a shift in the AI industry toward optimizing for task-completion costs rather than just raw model capability. By integrating directly with Cursor, xAI is targeting the specific needs of software engineers who require reliable, low-latency tools for complex coding tasks. This launch adds competitive pressure on established providers to lower prices and further demonstrates the growing trend of training models specifically for agentic, tool-using environments.