xAI Releases Grok 4.6
SpaceXAI launches Grok 4.6 with top benchmark scores, lower pricing than GPT‑5.6 and Fable 5, plus long‑horizon agent capabilities now live in Grok Bot, Cursor, and API.
“AI Disruption” Publication 10,400 Subscriptions 20% Discount Offer Link.
Musk brings Grok 4.6, putting SpaceXAI back on the major model leaderboard.
It has overtaken GPT-5.6 Sol and Fable 5 Max in benchmarks, while costing only $2 per million input tokens and $6 per million output tokens—significantly cheaper than both rivals.
On GDPVal-AA v2, which measures real-world work capability, Grok 4.6 scored the highest of all at 1753, leaving the others behind. It also ranked first on AA-Briefcase and Harvey LAB.
The new model is already available in Grok Build, Cursor, Grok Bot, and the API. Cursor users get double usage for the first week.
The key focus of this upgrade is long-horizon agent tasks: the model is designed to keep working continuously for extended periods without supervision and without dropping offline.
Just the day before, SpaceXAI released another product, Grok Bot—a team of agents that can log into various tools on their own and work around the clock. Grok Bot can now call Grok 4.6 directly.
