benchgap
DeepSeek · model

DeepSeek V3.1 benchmark scores

As of 2026-11-22, DeepSeek V3.1 (DeepSeek) has measured scores on 15 benchmarks and estimated scores on 54 more.

BenchmarkScoreSource
AA-GPQA Diamond73.5%measured
AA-HLE6.7%measured
AA-SciCode41.7%estimated ± 3.8 pp, high confidence
AIME2652.8%estimated ± 4.3 pp, medium confidence
Artificial Analysis Intelligence Index13.7%measured
CritPt0.0%measured
GDPval-AA0.1%estimated ± 11.0 pp, low confidence
GPQA Diamond75.1%estimated ± 1.2 pp, high confidence
HLE12.2%estimated ± 6.5 pp, medium confidence
IFBench54.0%estimated ± 6.6 pp, low confidence
AA-LCR47.0%measured
AA-Omniscience Accuracy23.1%measured
SciCode41.0%estimated ± 3.6 pp, high confidence
BrowseComp39.8%estimated ± 13.0 pp, low confidence
IFEval88.1%estimated ± 2.8 pp, high confidence
LongBench v251.9%estimated ± 5.5 pp, medium confidence
τ²-bench results34.8%measured
AA Coding Index34.1%estimated ± 5.3 pp, medium confidence
APEX-Agents-AA1.6%estimated ± 11.4 pp, low confidence
BioMysteryBench (human-difficult)0.0%estimated ± 6.8 pp, low confidence
SWE-bench Verified66.0%measured
BFCL v432.3%estimated ± 8.5 pp, low confidence
SWE-bench Pro43.0%estimated ± 4.5 pp, high confidence
AA Agentic Index3.5%estimated ± 12.7 pp, low confidence
MMLU-Pro83.7%measured
AA-IFBench37.8%measured
AIME 202466.3%measured
BBH98.1%estimated ± 2.8 pp, medium confidence
GPQA74.9%measured
MMLU90.4%estimated ± 0.9 pp, high confidence
Vals GPQA Diamond72.6%estimated ± 3.2 pp, high confidence
Vals LiveCodeBench66.1%estimated ± 10.0 pp, low confidence
Vals MMLU-Pro81.0%estimated ± 2.3 pp, high confidence
MATH-50095.6%estimated ± 1.2 pp, high confidence
MMMLU83.3%estimated ± 0.7 pp, medium confidence
AIME 202549.8%measured
AA MMLU-Pro84.6%estimated ± 0.8 pp, high confidence
ARC-AGI-135.3%estimated ± 6.7 pp, low confidence
ARC-AGI-20.0%estimated ± 10.9 pp, low confidence
FrontierCode 1.1 Main2.8%estimated ± 4.2 pp, low confidence
Mystery Game Puzzles10.9%estimated ± 4.7 pp, medium confidence
Vals SWE-bench57.3%estimated ± 6.6 pp, medium confidence
HealthBench Professional (length-adjusted, Artificial Analysis)50.0%estimated ± 4.3 pp, low confidence
HealthBench Professional (raw, Artificial Analysis)48.1%estimated ± 1.1 pp, low confidence
DeepSWE22.4%estimated ± 12.7 pp, low confidence
HLE w/o tools26.8%estimated ± 4.9 pp, high confidence
ARC-AGI-31.3%estimated ± 9.8 pp, low confidence
ARC-AGI-3 Provider Adapter2.7%estimated ± 7.3 pp, low confidence
CompileBench64.4%measured
Vibe Code Bench0.0%estimated ± 8.4 pp, medium confidence
C-Eval89.7%estimated ± 1.0 pp, medium confidence
HMMT Feb 202626.4%estimated ± 11.0 pp, low confidence
MMLU-Redux92.5%estimated ± 0.6 pp, high confidence
NL2Repo19.6%estimated ± 6.2 pp, low confidence
SuperGPQA54.8%estimated ± 8.5 pp, low confidence
HealthBench Hard0.0%estimated ± 8.6 pp, low confidence
MMLU-Pro (Arcee)77.1%estimated ± 4.0 pp, low confidence
React Native Evals73.3%estimated ± 2.7 pp, medium confidence
SWE-Rebench57.0%estimated ± 3.6 pp, low confidence
BioMysteryBench (human-solvable)81.7%estimated ± 0.6 pp, low confidence
HealthBench (length-adjusted)54.1%estimated ± 3.7 pp, medium confidence
HealthBench Professional56.7%estimated ± 4.2 pp, low confidence
HLE-Verified18.1%estimated ± 7.1 pp, low confidence
STEMBench2.4%estimated ± 0.2 pp, low confidence
SimpleQA22.4%estimated ± 4.3 pp, medium confidence
LiveCodeBench56.4%measured
MRCRv222.8%estimated ± 11.3 pp, low confidence
MMLongBench-Doc59.1%estimated ± 2.6 pp, low confidence
BioBasicsBench34.1%estimated ± 3.8 pp, low confidence