‹ PublicAI Index
The LLM benchmark aggregator.
DeepSeek V3.2
DeepSeek · 685B · open weights
Strongest in Tool use (#11 of 81), weakest in Harm refusal (#246 of 292). Above par in 14 of 26 scopes. Among the models it meets almost everywhere, it finishes behind Claude Opus 5 and Claude Fable 5 and ahead of GLM-4.6 and Gemini 2.5 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.
Agents53.5−12.9#75/2982/6
Tool use63.7−10.3#11/811/1
Knowledge work47.8−22.9#115/2081/2
Knowledge51.9−10.6#88/1971/3
Academic knowledge54.6−9.3#59/1231/1
Reasoning50.7−14.7#92/1952/5
Mathematics54.9−8.9#57/1521/2
Science52.9−10.6#67/1221/1
Reasoning42.7−26.3#102/1471/3
Coding48.1−18.1#98/1792/6
Agentic coding48−18.9#95/1732/5
Core abilities49.8−15.5#102/1991/4
General intelligence49.7−19.7#92/1781/3
Human preference58.9−9.9#105/3541/1
Human preference58.9−9.9#105/3541/1
Professional48.2−14.2#123/1941/2
Medical55.6−10.9#37/1491/1
Legal46.4−19.5#116/1731/1
Finance45.3−17.1#123/1621/1
Safety48.9−12.5#212/3292/3
Factual grounding59.2−11.5#27/1041/1
Secure code57.3−8.9#92/2701/1
Fairness50.5−19.5#104/2921/2
Toxicity avoidance51−8#184/2681/1
Jailbreak resistance36.8−28.2#227/2681/1
Harm refusal44.7−16.8#246/2921/2
§ 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 DeepSeek V3.2, left for the other.
§ 3 · Sources
Where the numbers come from
9 publications, 23 figures. Every one links to the page it was read from.
LMArena Text 1423
GDPval-AA 699
ARC-AGI-2 4%
Aider polyglot 74.2%
BFCL v4 56.73%
Kagi LLM Benchmark 52.2%
Vals · CaseLaw 55.41%Vals · LegalBench 76.08%Vals · CorpFin 50.97%Vals · TaxEval 68.15%Vals · MedQA 93.92%Vals · SWE-bench Verified 67.6% (Mini-SWE-agent)Vals · Vibe Code Bench 5.11% (OpenHands)Vals · GPQA Diamond 80.3%Vals · MMLU Pro 84.92%Vals · AIME 84.58%
Enkrypt · Jailbreak risk 23.9%Enkrypt · Harmful content risk 53.3%Enkrypt · CBRN risk 18.5%Enkrypt · Toxicity risk 5.5%Enkrypt · Bias risk 77%Enkrypt · Insecure code risk 18.2%
Vectara · Factual consistency 93.7%
Badge
[](https://publicai.io/model-index/m/deepseek-v3-2)