Inference
EN
xAI

Grok 3 [Beta]

Grok 3 [Beta] is a standard model from xAI in the Grok 3 family. 13 benchmarks count toward its index, in 4 categories.

  • Closed weights
  • Standard
  • Input: text
  • Released 2025-02
RankedEnough results for an index.
IndexAll categories, out of 100.39.3±8.5Median 51.0
CoverageIndex weight that has results.60%
SpeedOutput tokens per second.—Median 67/s
First tokenSeconds to the first output token.—Median 1.3 s
Input / 1MUS dollars per 1M input tokens.—Median $0.60
Output / 1MUS dollars per 1M output tokens.—Median $2.55
ContextMaximum tokens in one request.128KMedian 262K
EloLMArena rating and rank.—Median 1422

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.None counted——
CodingCode writing and repair.1 counted34.9—
ReasoningLogic problems and puzzles.2 counted34.4193 / 227
MultimodalTasks with images and text.None counted——
KnowledgeFacts and expert knowledge.3 counted47.1155 / 268
MultilingualTasks in many languages.None counted——
InstructionTasks with strict prompt rules.None counted——
MathMath problems.7 counted38.2128 / 168

Results

13 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
MATH level 5Math88.7% ±0.917 / 88
Measuring Short-Form Factuality in Large Language ModelsKnowledge43.6%3 / 14
MATH 500Math89.8% ±1.427 / 59
GPQA diamond · Epoch AIKnowledge75.8% ±2.7103 / 217
Massive Multitask Language Understanding ProfessionalKnowledge79.9%39 / 75
Graduate-Level Google-Proof Q&AKnowledge75.4%79 / 141
OTIS Mock AIME 2024-2025Math55.6% ±7.5116 / 200
LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for CodeCoding57.0%7 / 12
AIMEMath58.8% ±1.458 / 95
GPQA Diamond · Vals AIKnowledge74.2% ±2.285 / 135
MMLU ProKnowledge79.9% ±0.491 / 135
FrontierMath-2025-02-28-PrivateMath3.8% ±1.150 / 69
LiveCodeBenchCoding52.9% ±1.1114 / 142
American Invitational Mathematics Examination 2024Math52.2%22 / 27
FrontierMath-Tier-4-2025-07-01-PrivateMath0.0%45 / 55
ARC-AGI-2 (semi-private)Reasoning0.0%120 / 131
ARC-AGI-1 (semi-private)Reasoning5.5%119 / 126
TaxEval v2Finance75.9% ±0.810 / 143
FrontierMath-Tier-4-2025-07-01-PublicMath0.0%3 / 36
MGSMMultilingual91.3% ±1.728 / 74
LegalBenchLegal82.6% ±0.467 / 149
FrontierMath-2025-02-28-PublicMath0.0%33 / 64
CorpFin v2Finance59.7% ±1.074 / 131
MedQAHealthcare83.8% ±0.363 / 95
ARC-AGI-2 (public eval)Reasoning0.0%102 / 122
ARC-AGI-1 (public eval)Reasoning8.4%111 / 116

Sources

BenchLMCC BY-NC 4.0Vals AINo licence statedEpoch AICC BYARC Prize FoundationTerms restrict reuse

History

5 releases
Index40.4 → 39.3
23 Oct 202612 Nov 2026

Questions

How good is Grok 3 [Beta]?
Its index is 39.3 out of 100. That is place 211 of 283 ranked models.
What is Grok 3 [Beta] best at?
Knowledge: 47.1 out of 100, place 155 of 268.
What is the context window of Grok 3 [Beta]?
128K tokens.
Is Grok 3 [Beta] open source?
No. Its weights are not published.
When was Grok 3 [Beta] released?
February 2025.

Badge

Grok 3 [Beta] on Inference360: place 211 of 283, index 39.3
HTML
<a href="https://inference360.ai/models/grok-3-beta"><img src="https://inference360.ai/badges/grok-3-beta.svg" alt="Grok 3 [Beta] on Inference360: place 211 of 283, index 39.3" height="20"></a>
Markdown
[![Grok 3 \[Beta\] on Inference360: place 211 of 283, index 39.3](https://inference360.ai/badges/grok-3-beta.svg)](https://inference360.ai/models/grok-3-beta)

Compare with

Grok 3 Mini44.3Grok 450.0

More from xAI

All models
Grok 3 Mini44.3Grok 450.0Grok Code Fast 141.2Grok 229.1Grok 2 miniUnranked