benchgap
DeepSeek · model

DeepSeek V4 Pro benchmark scores

As of 2026-11-19, DeepSeek V4 Pro (DeepSeek) has measured scores on 50 benchmarks and estimated scores on 112 more.

BenchmarkScoreSource
AA-GPQA Diamond92.8%measured
AA-HLE41.0%measured
AA-SciCode51.0%measured
AIME2697.7%estimated ± 1.2 pp, medium confidence
Artificial Analysis Intelligence Index53.2%measured
CritPt18.0%measured
GDPval-AA54.5%measured
GPQA Diamond90.1%measured
HLE42.7%measured
IFBench74.9%estimated ± 6.6 pp, medium confidence
AA-LCR80.3%measured
LiveCodeBench v687.8%estimated ± 1.7 pp, medium confidence
AA-Omniscience Accuracy49.1%measured
SciCode51.7%estimated ± 2.9 pp, medium confidence
Terminal-Bench 2.187.9%measured
BrowseComp83.4%measured
HLE w/ tools60.0%measured
IFEval93.1%estimated ± 2.8 pp, high confidence
LongBench v265.9%estimated ± 5.5 pp, medium confidence
VITA-Bench45.8%estimated ± 6.7 pp, low confidence
τ²-bench results96.2%measured
AA Coding Index68.8%measured
APEX-Agents27.7%estimated ± 4.9 pp, low confidence
APEX-Agents-AA24.3%measured
BioMysteryBench (human-difficult)47.6%estimated ± 6.8 pp, low confidence
DeepSearchQA93.3%estimated ± 4.5 pp, high confidence
SWE-bench Verified80.6%measured
AutomationBench31.8%measured
BFCL v477.0%estimated ± 2.7 pp, medium confidence
CyberGym83.3%measured
JobBench37.4%estimated ± 7.3 pp, low confidence
SkillsBench55.2%estimated ± 4.1 pp, medium confidence
SWE-bench Pro55.4%measured
WideResearch83.0%estimated ± 1.6 pp, medium confidence
AA Agentic Index49.6%measured
MMLU-Pro87.5%measured
AA-IFBench76.5%measured
BBH100.0%estimated ± 2.8 pp, medium confidence
GPQA90.1%measured
MMLU90.5%estimated ± 0.9 pp, medium confidence
Vals GPQA Diamond92.4%measured
Vals LiveCodeBench87.5%measured
Vals MMLU-Pro87.0%measured
MMMLU90.6%estimated ± 0.7 pp, low confidence
Chess Puzzles41.8%estimated ± 10.8 pp, low confidence
AIME 202598.7%estimated ± 3.8 pp, low confidence
AA MMLU-Pro88.7%estimated ± 0.7 pp, medium confidence
Gert Labs69.2%estimated ± 5.5 pp, low confidence
AA-AnalystAgent43.0%estimated ± 8.0 pp, low confidence
AA EnterpriseOps-Gym49.6%measured
AA Harvey LAB v1.093.4%estimated ± 1.2 pp, medium confidence
AA Tau3 Banking39.6%measured
AA Terminal-Bench 2.180.0%estimated ± 2.5 pp, medium confidence
ApprenticeBench22.6%estimated ± 4.0 pp, high confidence
ARC-AGI-190.0%measured
ARC-AGI-261.3%measured
CursorBench 3.146.5%estimated ± 3.4 pp, low confidence
CursorBench 3.251.1%estimated ± 3.6 pp, high confidence
EBR-bench34.7%estimated ± 11.1 pp, low confidence
Terminal-Bench 3.011.5%estimated ± 3.2 pp, medium confidence
FrontierCode 1.1 Main45.8%estimated ± 3.5 pp, low confidence
FrontierMath v2 (Tier 4)26.8%measured
FrontierMath v2 (Tiers 1-3)64.6%measured
FrontierSWE v27.6%estimated ± 8.3 pp, medium confidence
Furniture Assembly38.2%estimated ± 12.4 pp, low confidence
MaxxBench51.9%estimated ± 5.7 pp, medium confidence
MazeBench2.7%estimated ± 4.0 pp, high confidence
MirrorCode33.3%measured
Mystery Game Puzzles29.0%estimated ± 13.3 pp, low confidence
OSWorld-Verified79.7%estimated ± 4.7 pp, high confidence
Terminal-Bench 4.16.7%measured
Terminal-Bench Hard54.3%estimated ± 2.8 pp, medium confidence
Vals SWE-bench96.4%measured
Terminal-Bench 2.1 (Vals AI)54.7%measured
VulcanBench v385.4%estimated ± 5.6 pp, low confidence
CWE-Bench-AA59.2%estimated ± 5.2 pp, medium confidence
AA Cyber Index29.0%estimated ± 9.2 pp, low confidence
DeepsecBench-AA19.2%estimated ± 5.2 pp, low confidence
HealthBench Professional (length-adjusted, Artificial Analysis)62.1%estimated ± 4.3 pp, medium confidence
HealthBench Professional (raw, Artificial Analysis)64.2%estimated ± 0.8 pp, low confidence
AA ITBench48.7%estimated ± 4.0 pp, high confidence
Bug Hunt Bench27.2%estimated ± 7.2 pp, low confidence
CursorBench 4.030.3%estimated ± 2.0 pp, medium confidence
CWE-bench v158.7%estimated ± 6.4 pp, medium confidence
DeepSWE62.7%measured
DRACO85.2%estimated ± 2.0 pp, medium confidence
FrontierCode 2.019.9%estimated ± 6.7 pp, low confidence
FrontierPhysics2.5%estimated ± 4.5 pp, high confidence
HLE w/o tools47.0%estimated ± 2.7 pp, medium confidence
OSWorld 2.019.4%estimated ± 5.5 pp, low confidence
OSWorld 2.169.7%estimated ± 6.0 pp, medium confidence
OSWorld 2.1 Offline70.0%estimated ± 1.9 pp, medium confidence
PostTrainBench v1.135.1%estimated ± 3.7 pp, high confidence
StealthMark42.8%measured
SWE Multilingual76.2%measured
Terminal-Bench 4.09.3%estimated ± 4.7 pp, medium confidence
Toolathlon-Verified74.1%measured
VulcanBench Frontier v481.6%estimated ± 4.0 pp, low confidence
AA AutomationBench42.4%estimated ± 7.9 pp, low confidence
GDP.pdf17.0%estimated ± 4.6 pp, high confidence
AA Terminal-Bench 4.06.4%estimated ± 6.3 pp, medium confidence
ARC-AGI-30.0%estimated ± 8.3 pp, low confidence
ARC-AGI-3 Provider Adapter2.7%estimated ± 7.3 pp, medium confidence
CWE-bench51.7%estimated ± 11.0 pp, low confidence
MKA41.1%estimated ± 6.4 pp, low confidence
Vibe Code Bench49.9%measured
FACTS Multimodal41.7%estimated ± 4.7 pp, high confidence
Hermes Bench70.1%estimated ± 1.5 pp, medium confidence
Hermes Index36.1%estimated ± 7.0 pp, low confidence
ExploitBench31.9%estimated ± 12.6 pp, low confidence
ExploitGym11.0%estimated ± 9.2 pp, low confidence
C-Eval92.8%estimated ± 1.2 pp, high confidence
Claw-Eval69.9%estimated ± 6.3 pp, low confidence
DeepPlanning47.0%estimated ± 13.1 pp, low confidence
HMMT Feb 202695.2%measured
HMMT Nov 202595.2%estimated ± 1.5 pp, low confidence
INCLUDE85.4%estimated ± 2.0 pp, medium confidence
MCP Atlas73.6%measured
MCP-Tasks77.5%estimated ± 7.0 pp, low confidence
MMAnswerBench90.3%estimated ± 2.4 pp, low confidence
MMLU-ProX85.3%estimated ± 1.3 pp, high confidence
MMLU-Redux94.5%estimated ± 0.6 pp, high confidence
NL2Repo61.5%measured
PostTrainBench v119.8%estimated ± 3.1 pp, medium confidence
QwenClawBench61.5%estimated ± 2.6 pp, high confidence
SuperGPQA82.4%estimated ± 7.2 pp, medium confidence
Terminal-Bench 2.067.9%measured
Toolathlon51.8%measured
AA LiveCodeBench100.0%estimated ± 2.6 pp, medium confidence
HealthBench Hard2.4%estimated ± 5.1 pp, low confidence
LiveCodeBench Pro80.0%estimated ± 6.0 pp, low confidence
MedXpertQA (Text)54.1%estimated ± 4.5 pp, medium confidence
MMLU-Pro (Arcee)86.9%estimated ± 3.4 pp, low confidence
React Native Evals79.9%estimated ± 2.7 pp, medium confidence
ResearchClawBench18.7%estimated ± 1.8 pp, high confidence
SWE-Rebench61.8%estimated ± 3.6 pp, medium confidence
PostTrainBench34.8%estimated ± 3.7 pp, high confidence
DAYJOB2.3%estimated ± 3.2 pp, high confidence
DAYJOB Healthcare1.1%estimated ± 4.7 pp, high confidence
BioMysteryBench (human-solvable)88.8%estimated ± 0.6 pp, medium confidence
FrontierCode 1.1 Extended61.1%estimated ± 2.4 pp, medium confidence
HealthBench (raw)49.6%estimated ± 6.0 pp, medium confidence
HealthBench (length-adjusted)55.1%estimated ± 2.5 pp, low confidence
HealthBench Professional51.3%estimated ± 4.2 pp, low confidence
HealthBench Professional (raw)60.2%estimated ± 4.5 pp, high confidence
HLE-Verified52.7%estimated ± 2.0 pp, medium confidence
LABBench281.5%estimated ± 1.8 pp, low confidence
ScarfBench63.8%estimated ± 3.8 pp, medium confidence
STEMBench2.6%estimated ± 0.2 pp, medium confidence
SimpleQA35.0%estimated ± 7.8 pp, low confidence
LiveCodeBench79.6%estimated ± 14.6 pp, low confidence
Agents' Last Exam25.7%measured
OpenHarmony Bench v1.059.0%measured
SEC-Bench Pro50.2%estimated ± 9.3 pp, medium confidence
IMOAnswerBench89.8%measured
MRCRv285.8%estimated ± 11.3 pp, low confidence
MMLongBench-Doc61.4%estimated ± 2.6 pp, low confidence
BioBasicsBench22.0%estimated ± 2.2 pp, low confidence
LifeVal74.8%estimated ± 1.6 pp, medium confidence
ExploitBench Internal Port0.0%estimated ± 6.8 pp, low confidence
PinchBench90.4%estimated ± 3.6 pp, low confidence
SWE-Atlas Codebase QnA57.9%estimated ± 4.8 pp, high confidence