Grok Ranking Update (December 16) Grok Code Fast 1 (Market Leader) This model is currently the absolute engine of the global AI agent economy, leading the leaderboard by a huge margin. OpenRouter ranks first overall on the leaderboard with 548 billion tokens and a 38% market share. It ranks first in all token categories, accounting for 28.4% of the market. It ranks first in all language token categories, with 138 billion tokens and an 11.3% market share. Kilo code leaderboard ranks first. Cline leaderboard ranks first. Roo code leaderboard ranks first. BLACKBOXAI leaderboard ranks first. It ranks second in tool usage, indicating widespread adoption by autonomous agents. It ranks third in the programming category, with 19.3 billion tokens. Grok 4.1 / 4.1 Fast (Agent and Emotional Intelligence Leader) The Grok 4.1 series excels in complex "perceived environment" scenarios, combining a context window of up to 2 million tokens with industry-leading emotional intelligence. It achieved a score of 1586 in the EQ-Bench3 test, setting a new record for emotional intelligence. Ranked #1 in the Tau-Squared Bench Telecom Smart Tools Usage Test. Ranked #2 in the LMArena Text Arena Test, Grok 4.1's mindset is second only to the Gemini 3 Pro. Ranked #2 in the Creative Writing v3 Test, outperforming Claude 4.5 and GPT-5.1. Ranked #5 in the Context Length Test, supporting 2 million token windows. Grok 4.20 (Frontier Inference Model) This internal xAI model is currently a top performer in Alpha Arena (a high-stakes real-world inference competition). Ranked #1 in the Alpha Arena Season 1.5 Test, achieving a 12.11% profit in autonomous stock trading. Ranked #1 in the Sentiment Arbitrage Test, outperforming GPT-5.2 and Gemini 3 using real-time X Firehose data.