SAN FRANCISCO — xAI is keeping its rapid release cadence going. The company behind Grok has launched Grok 4.6, a new frontier model built for long-running agents and more ambitious interactive and visual work. Grok 4.6 matches GPT-5.6 Sol on the Artificial Analysis Intelligence Index — a composite of nine benchmarks — placing it among the very best models in the world, as detailed in xAI's announcement. It is available immediately in Cursor and Grok Build.
Built to go the distance
Grok 4.6 builds on Grok 4.5 with a focus on staying with complex tasks across many steps, whether researching a topic, analyzing information, working across a codebase, or turning an idea into a polished application. On longer trajectories, xAI says the model increasingly self-tests and verifies its own work before moving on — a hallmark of more capable, agentic AI.
The model achieves frontier intelligence across several agentic coding and knowledge-work benchmarks. It posts a 61 on the AA Intelligence Index, level with GPT-5.6 Sol, and improves on Grok 4.5 across the board, including a jump to 65.9% on DeepSWE 1.1 and 69.9% on CursorBench 3.2. Those gains reflect the pace at which xAI, now part of SpaceX after its merger, keeps closing ground on and overtaking rivals.
Strong on real-world projects
xAI tested Grok 4.6 on projects designed to stretch its range, and found it especially strong at turning a broad product idea into a working first version. The model can research an unfamiliar domain, structure an application, implement the core interactions, and keep refining through several rounds of feedback. It also produces stronger first passes on visual and interactive projects than Grok 4.5, establishing structure and visual language in a single pass. That momentum follows recent releases like Grok Imagine Image 2.0, which sharpened the platform's creative tools.
