benchgap
DeepSeek · model

DeepSeek V4 Flash 0731 benchmark scores

As of 2026-11-22, DeepSeek V4 Flash 0731 (DeepSeek) has measured scores on 47 benchmarks and estimated scores on 119 more.

BenchmarkScoreSource
AA-GPQA Diamond90.8%measured
AA-HLE38.6%measured
AA-SciCode50.3%measured
AIME2696.7%estimated ± 1.3 pp, medium confidence
Artificial Analysis Intelligence Index34.3%measured
CritPt16.6%measured
GDPval-AA46.7%measured
GPQA Diamond88.1%measured
HLE34.8%measured
AA-LCR79.7%measured
LiveCodeBench v685.5%estimated ± 1.7 pp, medium confidence
AA-Omniscience Accuracy40.4%measured
SciCode51.3%estimated ± 2.9 pp, medium confidence
Terminal-Bench 2.182.7%measured
BrowseComp73.2%measured
HLE w/ tools45.1%measured
LongBench v265.4%estimated ± 5.5 pp, medium confidence
VITA-Bench42.1%estimated ± 6.7 pp, low confidence
τ²-bench results94.6%estimated ± 3.1 pp, high confidence
AA Coding Index69.1%measured
APEX-Agents37.5%estimated ± 7.4 pp, low confidence
APEX-Agents-AA19.3%estimated ± 4.4 pp, high confidence
BioMysteryBench (human-difficult)47.1%estimated ± 6.8 pp, low confidence
DeepSearchQA92.1%estimated ± 4.5 pp, high confidence
SWE-bench Verified79.0%measured
AutomationBench25.1%measured
BFCL v476.4%estimated ± 2.7 pp, medium confidence
CyberGym76.7%measured
JobBench35.0%estimated ± 7.3 pp, low confidence
SkillsBench44.5%estimated ± 4.1 pp, medium confidence
SWE-bench Pro52.6%measured
WideResearch82.7%estimated ± 1.6 pp, medium confidence
AA Agentic Index41.7%measured
MMLU-Pro86.2%measured
BBH99.5%estimated ± 2.8 pp, medium confidence
FailureSensorIQ65.8%estimated ± 2.1 pp, high confidence
GPQA88.1%measured
MMLU90.5%estimated ± 0.9 pp, medium confidence
Vals GPQA Diamond89.9%measured
Vals LiveCodeBench87.3%measured
Vals MMLU-Pro86.2%measured
MMMLU90.4%estimated ± 0.7 pp, low confidence
Chess Puzzles32.0%measured
AIME 202598.6%estimated ± 3.8 pp, low confidence
AA MMLU-Pro87.4%estimated ± 0.7 pp, high confidence
Gert Labs46.8%measured
AA-AnalystAgent46.5%estimated ± 3.7 pp, low confidence
AA Harvey LAB v1.092.5%estimated ± 1.2 pp, medium confidence
AA Tau3 Banking39.4%measured
AA Terminal-Bench 2.181.9%estimated ± 2.5 pp, high confidence
ApprenticeBench7.0%estimated ± 4.0 pp, high confidence
ARC-AGI-189.0%measured
ARC-AGI-261.4%measured
BabaIsBench75.0%measured
CursorBench 3.144.7%estimated ± 3.4 pp, low confidence
CursorBench 3.267.8%estimated ± 3.3 pp, medium confidence
EBR-bench24.5%estimated ± 10.4 pp, low confidence
Terminal-Bench 3.010.6%estimated ± 6.9 pp, low confidence
FrontierCode 1.1 Main41.0%estimated ± 3.5 pp, low confidence
FrontierMath v2 (Tier 4)29.4%estimated ± 4.1 pp, medium confidence
FrontierMath v2 (Tiers 1-3)54.8%estimated ± 9.0 pp, medium confidence
FrontierSWE v22.8%estimated ± 8.3 pp, low confidence
Furniture Assembly34.5%estimated ± 9.3 pp, medium confidence
MaxxBench28.1%measured
MazeBench0.0%measured
Mystery Game Puzzles34.0%measured
NYT Connections68.9%estimated ± 10.2 pp, low confidence
OSWorld-Verified78.2%estimated ± 4.7 pp, high confidence
Terminal-Bench 4.119.7%estimated ± 10.9 pp, low confidence
Terminal-Bench Hard51.5%estimated ± 3.9 pp, high confidence
Vals SWE-bench88.8%measured
Terminal-Bench 2.1 (Vals AI)67.0%measured
VulcanBench v388.4%measured
CWE-Bench-AA43.3%estimated ± 5.2 pp, medium confidence
AA Cyber Index25.3%estimated ± 9.2 pp, low confidence
DeepsecBench-AA15.3%estimated ± 5.2 pp, low confidence
HealthBench Professional (length-adjusted, Artificial Analysis)56.3%estimated ± 4.3 pp, low confidence
HealthBench Professional (raw, Artificial Analysis)63.1%estimated ± 0.8 pp, low confidence
AA ITBench43.5%estimated ± 3.2 pp, high confidence
Bug Hunt Bench19.5%estimated ± 7.2 pp, low confidence
CursorBench 4.030.2%estimated ± 2.0 pp, medium confidence
CWE-bench v157.4%estimated ± 5.1 pp, medium confidence
DeepSWE54.4%measured
DRACO86.9%estimated ± 0.7 pp, low confidence
FrontierCode 2.019.4%estimated ± 6.7 pp, low confidence
FrontierPhysics2.9%estimated ± 5.0 pp, medium confidence
HLE w/o tools38.5%estimated ± 2.7 pp, high confidence
OSWorld 2.019.4%estimated ± 5.5 pp, low confidence
OSWorld 2.156.6%estimated ± 3.6 pp, medium confidence
OSWorld 2.1 Offline50.0%estimated ± 1.9 pp, medium confidence
PostTrainBench v1.130.8%estimated ± 3.7 pp, high confidence
SpireClimb0.2%estimated ± 9.3 pp, medium confidence
StealthMark20.6%estimated ± 7.7 pp, low confidence
SWE Multilingual73.3%measured
Toolathlon-Verified70.3%measured
VulcanBench Frontier v481.7%estimated ± 4.0 pp, low confidence
AA AutomationBench41.9%estimated ± 7.9 pp, low confidence
GDP.pdf18.9%estimated ± 4.2 pp, high confidence
AA Terminal-Bench 4.04.4%estimated ± 11.0 pp, low confidence
ARC-AGI-30.0%estimated ± 8.3 pp, low confidence
ARC-AGI-3 Provider Adapter2.7%estimated ± 7.3 pp, medium confidence
cursorBench4144.8%estimated ± 3.7 pp, medium confidence
CWE-bench43.5%estimated ± 4.6 pp, low confidence
MKA34.8%estimated ± 4.6 pp, high confidence
SkillsBench official leaderboard (BenchFlow)54.2%estimated ± 6.2 pp, medium confidence
CompileBench100.0%estimated ± 10.0 pp, low confidence
Vibe Code Bench34.1%estimated ± 8.4 pp, medium confidence
FACTS Multimodal40.6%estimated ± 5.6 pp, medium confidence
Hermes Bench70.0%estimated ± 1.5 pp, medium confidence
Hermes Index37.2%estimated ± 3.1 pp, high confidence
ExploitBench55.3%estimated ± 7.2 pp, medium confidence
ExploitGym11.9%estimated ± 2.7 pp, high confidence
C-Eval94.3%estimated ± 0.9 pp, low confidence
Claw-Eval68.5%estimated ± 6.3 pp, low confidence
DeepPlanning44.7%estimated ± 13.1 pp, low confidence
HMMT Feb 202694.8%measured
HMMT Nov 202596.2%estimated ± 2.8 pp, medium confidence
INCLUDE84.6%estimated ± 2.0 pp, medium confidence
MCP Atlas69.0%measured
MCP-Tasks75.4%estimated ± 7.0 pp, low confidence
MMAnswerBench87.6%estimated ± 2.8 pp, low confidence
MMLU-ProX84.3%estimated ± 1.3 pp, high confidence
MMLU-Redux93.8%estimated ± 0.6 pp, high confidence
NL2Repo54.2%measured
PostTrainBench v135.3%estimated ± 3.6 pp, medium confidence
QwenClawBench59.1%estimated ± 2.6 pp, high confidence
SuperGPQA75.0%estimated ± 7.2 pp, medium confidence
Terminal-Bench 2.056.9%measured
Toolathlon47.8%measured
AA LiveCodeBench97.0%estimated ± 2.6 pp, medium confidence
HealthBench Hard0.0%estimated ± 5.1 pp, low confidence
LiveCodeBench Pro77.7%estimated ± 6.0 pp, low confidence
MedXpertQA (Text)48.6%estimated ± 4.5 pp, low confidence
MMLU-Pro (Arcee)86.2%estimated ± 3.4 pp, medium confidence
React Native Evals79.2%estimated ± 2.7 pp, medium confidence
ResearchClawBench18.2%estimated ± 1.8 pp, high confidence
SWE-Rebench61.7%estimated ± 3.6 pp, medium confidence
PostTrainBench29.3%estimated ± 3.7 pp, high confidence
DAYJOB1.3%estimated ± 3.2 pp, high confidence
DAYJOB Healthcare0.7%estimated ± 4.2 pp, high confidence
Finance Agent v255.6%estimated ± 2.4 pp, medium confidence
BioMysteryBench (human-solvable)88.3%estimated ± 0.6 pp, medium confidence
FrontierCode 1.1 Extended53.3%estimated ± 3.0 pp, low confidence
HealthBench (raw)47.4%estimated ± 6.0 pp, medium confidence
HealthBench (length-adjusted)54.7%estimated ± 2.5 pp, low confidence
HealthBench Professional47.4%estimated ± 4.2 pp, low confidence
HealthBench Professional (raw)58.1%estimated ± 4.5 pp, high confidence
HLE-Verified25.8%estimated ± 2.0 pp, low confidence
LABBench281.0%estimated ± 1.8 pp, low confidence
ScarfBench61.9%estimated ± 3.8 pp, low confidence
BlenderWorld (text-only)71.8%estimated ± 2.8 pp, medium confidence
Cookie-Bench81.0%estimated ± 1.3 pp, medium confidence
STEMBench2.5%estimated ± 0.2 pp, medium confidence
SimpleQA46.1%estimated ± 4.3 pp, low confidence
LiveCodeBench78.7%estimated ± 14.6 pp, low confidence
Agents' Last Exam25.2%measured
OpenHarmony Bench v1.053.8%measured
SEC-Bench Pro51.8%estimated ± 9.3 pp, medium confidence
IMOAnswerBench88.4%measured
MRCRv285.7%estimated ± 11.3 pp, low confidence
MMLongBench-Doc61.4%estimated ± 2.6 pp, low confidence
BioBasicsBench33.9%estimated ± 2.8 pp, medium confidence
LifeVal62.6%estimated ± 1.6 pp, medium confidence
ExploitBench Internal Port0.1%estimated ± 1.4 pp, low confidence
PinchBench89.8%estimated ± 3.6 pp, medium confidence
SWE-Atlas Codebase QnA51.5%estimated ± 4.8 pp, high confidence