XNews

Grok 4.1 Release and Benchmark/Ranking Discussion

Grok 4.1 Release and Benchmark/Ranking Discussion

Grok 4.1 achieved multiple metrics and ranking improvements during its silent release and benchmarking phase.

Key insights

Grok 4.1 Achieves Top Ranking and High Elo on LM Arena

Official or test data shows that Grok 4.1 ranks first on the LM Arena leaderboard and achieves a score of 1483 Elo in the Overall Style Control category, indicating significant improvements in style control and conversational performance. This performance metric is supported by multiple test charts and ranking screenshots.

User Preference Rate During Silent Release at 64.78%

During the silent release period from November 1 to 14, 2025, Grok 4.1's preference rate was reported at 64.78%. This metric measures user or evaluation set preference for the new version and serves as a performance reference during the silent release.

Close to Polaris Alpha in Creative Writing Benchmark

In the Creative Writing v3 benchmark tests, Grok 4.1 and its Thinking variant ranked high, only slightly below Polaris Alpha (described as early GPT-5.1), demonstrating its competitiveness in creative writing tasks.

Source posts