‹ PublicAI Index
The LLM benchmark aggregator.
Grok 3 Mini
SpaceXAI
Strongest in Safe-prompt compliance (#7 of 82), weakest in Human preference (#164 of 354). Above par in 19 of 21 scopes. Among the models it meets almost everywhere, it finishes behind Claude Opus 4.8 and GPT-5.5 and ahead of O4 Mini and DeepSeek V4 Flash.
§ 1 · Profile
What it is good at
Bars run from 50 — the average of the models each source lists — so right of the line is above par. The middle column is the gap to whoever leads that scope.
Coding53.1−13.1#58/1792/6
Code generation56.8−8.5#9/201/1
Agentic coding51.3−15.6#77/1731/5
Professional53−9.4#61/1941/2
Legal53.7−12.2#58/1731/1
Medical53.5−13#61/1491/1
Finance53.3−9.1#72/1621/1
Core abilities52.9−12.4#68/1991/4
General intelligence54.2−15.2#56/1781/3
Reasoning50.3−15.1#97/1952/5
Mathematics55−8.8#56/1521/2
Science52.2−11.3#70/1221/1
Reasoning41.9−27.1#111/1471/3
Safety52.7−8.7#107/3291/3
Safe-prompt compliance59.5−0.9#7/821/1
Fairness56.9−13.1#50/2921/2
Harm refusal49.9−11.6#154/2921/2
Knowledge50.1−12.4#111/1971/3
Academic knowledge50.2−13.7#77/1231/1
Human preference53.2−15.6#164/3541/1
Human preference53.2−15.6#164/3541/1
§ 2 · Head to head
What it beats, and what beats it
The same models turn up scope after scope. Counted once: where both were placed, who finished higher. Bars run right for Grok 3 Mini, left for the other.
§ 3 · Sources
Where the numbers come from
7 publications, 17 figures. Every one links to the page it was read from.
LMArena Text 1364
ARC-AGI-2 0.4%
LiveCodeBench 66.7%
Aider polyglot 49.3%
Kagi LLM Benchmark 61.3%
Vals · LegalBench 83.14%Vals · CorpFin 61.11%Vals · TaxEval 72.98%Vals · MedQA 90.1%Vals · GPQA Diamond 79.29%Vals · MMLU Pro 81.37%Vals · AIME 85%
HELM Safety · HarmBench 57.2%HELM Safety · SimpleSafetyTests 99.3%HELM Safety · Anthropic Red Team 99.5%HELM Safety · BBQ 96.7%HELM Safety · XSTest 98.4%
Badge
[](https://publicai.io/model-index/m/grok-3-mini)