‹ PublicAI Index
The LLM benchmark aggregator.
GPT-4.1 nano
OpenAI
Strongest in Safe-prompt compliance (#39 of 82), weakest in Agents (#257 of 298). Above par in 6 of 26 scopes. Among the models it meets almost everywhere, it finishes behind GPT-5 and GPT-5.2 and ahead of Gemma 4 31B and Command A.
§ 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.
Safety52.8−8.6#105/3272/3
Safe-prompt compliance54.2−6.2#39/821/1
Secure code60.1−6.1#64/2681/1
Harm refusal53.8−7.7#91/2902/2
Jailbreak resistance53.7−11.3#133/2661/1
Toxicity avoidance53.3−5.7#145/2661/1
Fairness45.7−24.3#176/2902/2
Coding42.5−23.6#150/1791/6
Agentic coding41.8−24.9#147/1731/5
Core abilities43.4−22.2#161/2001/4
General intelligence40.4−28.6#150/1801/3
Knowledge40.8−21.7#176/1971/3
Academic knowledge27.8−36.1#118/1231/1
Reasoning38.1−27.3#179/1952/5
Science32.4−31.1#108/1221/1
Reasoning41.9−27.1#112/1471/3
Mathematics39.4−24.4#122/1521/2
Professional39.6−22.8#182/1941/2
Medical41.8−24.7#128/1491/1
Finance37.4−25#148/1621/1
Legal36.9−29#160/1731/1
Human preference49.2−19.6#206/3541/1
Human preference49.2−19.6#206/3541/1
Agents42.5−23.8#257/2982/6
Tool use46.5−27.5#41/811/1
Knowledge work32.8−37.9#196/2081/2
§ 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 GPT-4.1 nano, left for the other.
§ 3 · Sources
Where the numbers come from
9 publications, 25 figures. Every one links to the page it was read from.
LMArena Text 1322
GDPval-AA -206
ARC-AGI-2 0%
Aider polyglot 8.9%
BFCL v4 33.05%
Kagi LLM Benchmark 33.3%
Vals · LegalBench 61.06%Vals · CorpFin 42.08%Vals · TaxEval 60.75%Vals · MortgageTax 52.82%Vals · MedQA 68.22%Vals · GPQA Diamond 50.76%Vals · MMLU Pro 63.48%Vals · AIME 26.46%
HELM Safety · HarmBench 86.8%HELM Safety · SimpleSafetyTests 99%HELM Safety · Anthropic Red Team 99.6%HELM Safety · BBQ 87.5%HELM Safety · XSTest 96%
Enkrypt · Jailbreak risk 9.7%Enkrypt · Harmful content risk 40%Enkrypt · CBRN risk 10.5%Enkrypt · Toxicity risk 3.9%Enkrypt · Bias risk 87.1%Enkrypt · Insecure code risk 12.4%
Badge
[](https://publicai.io/model-index/m/gpt-4-1-nano)