Alex Dudov
⬤ Grok Code has jumped to #1 on the Kilo Agentic Model Leaderboard based on actual developer activity, not lab tests. The coding AI racked up 35.6 billion tokens of usage, pulling way ahead of everything else in the rankings. These numbers come from over 750,000 developers using Kilo Coders inside VS Code, so it's real work happening in real dev environments.
⬤ The gap between Grok Code and second place is massive—usage levels are now over four times higher than the runner-up. The leaderboard tracks token-level activity, which shows not just how many people are trying it out, but how heavily they're leaning on it day-to-day for actual coding work.
⬤ This kind of "real-world dev love" points to how fast AI coding assistants are getting adopted across the industry. Agentic models that can reason through problems, plan out solutions, and execute structured tasks are catching on hard, and Grok Code's surge shows serious traction with developers who are actually building stuff.
⬤ The spike in usage signals a shift from basic autocomplete suggestions to AI that handles full tasks and automation right inside your IDE. With deployment expanding this fast, agentic coding platforms are clearly becoming a major piece of where AI-powered dev tools are headed next.
Alex Dudov
