Benchmark Almanac

Grok 4.1

Grok 4.1 places in the top three on 2 benchmarks.

xAI Released November 17th, 2025 Family Grok 4.1 The lab's model card Compare with others
23
Benchmarks with a score
0
Ranked first, out of five or more models
2
In the top three
xAI Nov 2025

Grok 4.1

Closed weights Flagship

No Epoch Capabilities Index score yet
  • Virology Capabilities Test (VCT) 61% #4 /31
  • Cybench 39% #3 /16
  • LAB-Bench FigQA 34% #11 /11

Where it ranks

Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.

Knowledge and language

BenchmarkScoreRankSource
LMArena (Elo) 1,465 #45 of 335 Official leaderboard
xAI Factuality (Hallucination) eval 4.22% #1 of 2 Official leaderboard
Role-play Benchmark (MiniMax) 48.5 #8 of 8 Official leaderboard
EQ-Bench 3 1,586 #1 of 1 Lab model card

Math and science

BenchmarkScoreRankSource
ResearchClawBench 13.5 #20 of 23 Official leaderboard

Coding

BenchmarkScoreRankSource
Code Arena 1,217 #122 of 128 Official leaderboard
WebDev Arena 1,211 #102 of 105 arena.ai, via Epoch AI

Professional work

BenchmarkScoreRankSource
prinzbench 25.0 #18 of 26 Official leaderboard
APEX-1 57.5% #58 of 71 Official leaderboard

Safety and security

BenchmarkScoreRankSource
Virology Capabilities Test (VCT) 61% #4 of 31 Lab model card
Cybench 39% #3 of 16 Official leaderboard
AgentDojo 5.00% #4 of 8 Lab model card
AgentHarm 14% #3 of 6 Lab model card
BioLP-bench 37% #5 of 9 Lab model card
LAB-Bench ProtocolQA 79% #9 of 16 Lab model card
LAB-Bench CloningScenarios 46% #7 of 12 Lab model card
MASK 51% #17 of 19 Lab model card
LAB-Bench FigQA 34% #11 of 11 Lab model card
MakeMeSay 0% #10 of 10 Lab model card
General refusals (xAI) 7.00% #8 of 8 Lab model card
Refusals + System Jailbreak (xAI) 2.00% #5 of 5 Lab model card
Refusals + User Jailbreak (xAI) 2.00% #5 of 5 Lab model card
SycophancyEval (answer sycophancy) 19% #4 of 4 Lab model card

Its lab's numbers next to 2 leaderboards

Where the benchmark's own leaderboard, or Epoch AI's run, also has this model. Gaps mostly come from different harnesses, effort levels and task subsets. Every pair.

Benchmark Lab says Board says Gap What differs
LMArena (Elo) 1,483 LMArena Text Leaderboard overall Elo (Grok 4.1 Thinking, 'quasarflux') 1,465 Leaderboard, thinking +18 xAI quotes Grok 4.1 Thinking's launch rating, while the board reports its later Arena rating.
Cybench 39% CyBench, unguided success rate 0.39, Grok 4.1 Thinking (Table 4) 39% Leaderboard Same

Its launch post

The benchmarks the lab chose to headline, with the scores it gave.

Its system card

Every result in the longer documents the lab published with it. 15 of them aren't in the launch post.