Grok-3 mini

Grok-3 mini ranks first on 2 of the 67 benchmarks with five or more models scored.

xAI Released April 9th, 2025 Compare with others

Benchmarks with a score

Ranked first, out of five or more models

In the top three

xAI Apr 2025

Grok-3 mini

Closed weights

139.9 #126 of 256

Epoch Capabilities Index

Self-Reported Index 138.7 #187 of 404

GPQA Diamond 76.3% #107 /229
SimpleQA Verified 20.2% #41 /56
FrontierMath 5.86% #38 /55

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

Benchmark	Score	Rank	Source
IFEval	95.1%	#1 of 60	Official leaderboard
LLM Confabulation Leaderboard	10.8%	#4 of 58	Official leaderboard
Vectara Hallucination Leaderboard (HHEM)	3.30%	#52 of 148	Official leaderboard
LOFT	83.1%	#3 of 8	Lab model card
Fiction.LiveBench	66.7%	#24 of 57	Fiction.live leaderboard, via Epoch AI
LMArena (Elo)	1,364	#166 of 335	Official leaderboard
MMLU-Pro	78.9%	#88 of 167	Lab model card
ECLeKTic	17.1%	#20 of 35	Official leaderboard
FACTS Grounding	75%	#22 of 34	Official leaderboard
SimpleQA	21.7%	#39 of 60	Lab model card
EQ-Bench 3	1,088	#57 of 78	Official leaderboard
SimpleQA Verified	20.2%	#41 of 56	Official leaderboard
Phare	60%	#51 of 62	Official leaderboard
Global MMLU Lite	67.3%	#40 of 43	Official leaderboard

Math and science

Benchmark	Score	Rank	Source
Misguided Attention	73.7%	#1 of 45	Official leaderboard
AIME 2024	95.8%	#3 of 88	Lab model card
MATH Level 5	90.9%	#16 of 94	Epoch AI run
GPQA Main	67.5%	#14 of 60	Official leaderboard
MATH-500	94.2%	#10 of 38	Official leaderboard
AIME 2025	90.8%	#46 of 143	Lab model card
IneqMath	6.00%	#14 of 34	Official leaderboard
OTIS Mock AIME 2024-2025	77.8%	#99 of 212	Epoch AI run
GPQA Diamond	76.3%	#107 of 229	Epoch AI run
PolyMath	32.3%	#19 of 38	Official leaderboard
HMMT 2025	74.2%	#29 of 53	Official leaderboard
GSM8K Platinum Initial 268	98.5%	#12 of 19	Official leaderboard
FrontierMath	5.86%	#38 of 55	Official leaderboard
BRUMO 2025	85%	#32 of 42	Official leaderboard
SMT 2025	78.8%	#31 of 40	Official leaderboard
CMIMC 2025	68.1%	#28 of 33	Official leaderboard
ARC-AGI	16.5%	#82 of 95	Official leaderboard
ARC-AGI-2	0.42%	#88 of 98	Official leaderboard
PutnamBench	0%	#12 of 12	Official leaderboard

Coding

Benchmark	Score	Rank	Source
BigCodeBench-Hard	31.1%	#16 of 96	Official leaderboard
LiveCodeBench	82.9%	#10 of 21	Official leaderboard
LiveCodeBench v6	66.7%	#10 of 21	Official leaderboard
Aider Polyglot	49.3%	#22 of 46	Official leaderboard
LiveCodeBench v5	68.4%	#11 of 21	Official leaderboard
WeirdML	42.6%	#69 of 112	Official leaderboard
DesignArena Website	922	#140 of 151	Official leaderboard
DesignArena 3D Design	938	#128 of 136	Official leaderboard
DesignArena Game Dev	940	#135 of 143	Official leaderboard
DesignArena Code	921	#138 of 143	Official leaderboard
DesignArena UI Component	881	#135 of 139	Official leaderboard
DesignArena Data Visualization	875	#118 of 118	Official leaderboard

Agents and tools

Benchmark	Score	Rank	Source
Elimination Game	6	#6 of 56	Official leaderboard
Step Race	3	#16 of 59	Official leaderboard
LLM Chess	430	#46 of 123	Official leaderboard
LisanBench	849.3	#59 of 133	Official leaderboard
lmgame-Bench	29.0	#11 of 23	Official leaderboard
TextQuests	17.8	#15 of 28	Official leaderboard
Berkeley Function Calling Leaderboard (BFCL)	53%	#41 of 67	Official leaderboard

Professional work

Benchmark	Score	Rank	Source
TaxEval v2	73%	#54 of 129	Official leaderboard
LegalBench	83.1%	#58 of 138	Official leaderboard
Vals Finance Agent v1	31.7%	#22 of 47	Official leaderboard
CorpFin v2	61.1%	#59 of 119	Official leaderboard

Vision, audio and robotics

Benchmark	Score		Rank	Source
EgoSchema	74.3%		#3 of 12	Lab model card
MMMU	69.4%		#65 of 103	Lab model card

Safety and security

Benchmark	Score	Rank	Source
Anthropic Red Team	99.5%	#18 of 75	Official leaderboard
SimpleSafetyTests	99.3%	#30 of 75	Official leaderboard
Human Pathogen Capabilities Test v2 (HPCT v2)	51.6%	#39 of 71	Official leaderboard
HELM Safety	90.2	#46 of 75	Official leaderboard
AIR-Bench 2024	53.5%	#60 of 82	Official leaderboard
Human Pathogen Capabilities Test	40.9%	#69 of 86	Official leaderboard
LiveSecBench	28.4	#29 of 34	Official leaderboard
Molecular Biology Capabilities Test	37.5%	#82 of 90	Official leaderboard

Other

Benchmark	Score		Rank	Source
HELM Capabilities	67.9		#24 of 60	Official leaderboard

Its lab's numbers put it at 138.7, 1.2 points below its ECI

The Self-Reported Index fits one score to the 8 capability results xAI reported for this model, on the scale of the Epoch Capabilities Index. Labs choose which benchmarks to show and how to run them, so read it as the lab's claim, not an independent measurement. The 90% interval runs from 135.0 to 142.3. How we compute it.

AIME 2024 95.8
AIME 2025 90.8
GPQA Diamond 84
LOFT 83.1
MMLU-Pro 78.9
EgoSchema 74.3
MMMU 69.4
SimpleQA 21.7

Its lab's numbers next to 3 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
AIME 2025	90.8% AIME'25 (cons@64; pass@1 = 82.0) [Grok 3 mini Think]	81.7% Leaderboard, high	+9.1 pts
GPQA Diamond	84% GPQA (cons@64; pass@1 = 80.3) [Grok 3 mini Think]	76.3% Epoch AI run, low	+7.7 pts
MMLU-Pro	78.9% MMLU-pro [Grok 3 mini, non-reasoning table]	83% Leaderboard	−4.1 pts

xAI's models over time

xAI's models in color, against every other model on the Epoch Capabilities Index, by release date.

Source: Epoch Capabilities Index, Epoch AI (CC BY 4.0)

Its launch post

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

Grok 3 Beta: The Age of Reasoning Agents (xAI news) Feb 19, 2025