Model launches / DeepSeek

DeepSeek V4.1 Flash 1017

Scores at a glance

66 benchmarks · large dot: DeepSeek V4.1 Flash 1017; small dots: closest models where reported

Artificial Analysis Intelligence Index v5.0 ↗32.49 index

DeepSeek V4.1 Flash 1017 (Max) · Intelligence Index v5.0 ↗

1 more reported result

45.10 index · DeepSeek V4.1 Flash 1017 (Max) · Intelligence Index v4.3.2 · Artificial Analysis ↗

AA-Briefcase v1.1 ↗1570.42 Elo

DeepSeek V4.1 Flash 1017 (Max) · AA-Briefcase v1.1 · overall Elo ↗

1 more reported result

1570 Elo · DeepSeek-V4.1-Flash-1017 (max) · AA-Briefcase v1.1 · AA-Briefcase v1.1 (Elo-score) · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

AutomationBench-AA ↗66.64%

DeepSeek V4.1 Flash 1017 (Max) · AutomationBench-AA · guardrail-gated objective score ↗

1 more reported result

66.6% · DeepSeek-V4.1-Flash-1017 (max) · AutomationBench-AA · Automation Bench-AA · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

GDP.pdf ↗17.20%

DeepSeek V4.1 Flash 1017 (Max) · GDP.pdf · all-pass ↗

2 more reported results

73% · DeepSeek-V4.1-Flash-1017 (max) · GDP.pdf · GDP.pdf w/ python · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

17.2% · DeepSeek-V4.1-Flash-1017 (max) · GDP.pdf · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

Terminal-Bench 4.0 ↗33.84%

DeepSeek V4.1 Flash 1017 (Max) · Terminal-Bench 4.0 · AA harness ↗

3 more reported results

31.82 % · DeepSeek V4.1 Flash 1017 (Mini-SWE-agent) · Terminal-Bench 4.0 · Score · Vals AI · Terminal-Bench 4.0 ↗

42.3 · DS-V4.1-Flash-1017 · Terminal-Bench 4.0 · DeepSeek V4.1 Flash 1017 model card ↗

42.3% · DeepSeek-V4.1-Flash-1017 (max) · Terminal-Bench 4.0 · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

Terminal-Bench-Science 0.1 ↗12.86%

DeepSeek V4.1 Flash 1017 (Max) · Terminal-Bench-Science 0.1 · AA harness ↗

1 more reported result

15.7% · DeepSeek-V4.1-Flash-1017 (max) · Terminal-Bench-Science 0.1 · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

FrontierCode 2.0 ↗28.47%

DeepSeek V4.1 Flash 1017 (Max) · FrontierCode 2.0 · AA implementation ↗

CritPt v2 · hard subset ↗24.00%

DeepSeek V4.1 Flash 1017 (Max) · CritPt v2 · hard subset · AA implementation ↗

HLE-Diamond ↗25.30%

DeepSeek V4.1 Flash 1017 (Max) · HLE-Diamond · AA implementation ↗

1 more reported result

21.6% · DeepSeek-V4.1-Flash-1017 (max) · HLE-Diamond · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

AA-Omniscience v1.1 ↗0.90 index

DeepSeek V4.1 Flash 1017 (Max) · AA-Omniscience v1.1 · error-penalized index ↗

1 more reported result

4.70 index · DeepSeek V4.1 Flash 1017 (Max) · AA-Omniscience · error-penalized score · Artificial Analysis ↗

AA-Omniscience v1.1 · accuracy ↗39.60%

DeepSeek V4.1 Flash 1017 (Max) · AA-Omniscience v1.1 · accuracy ↗

1 more reported result

42.85% · DeepSeek V4.1 Flash 1017 (Max) · AA-Omniscience · accuracy · Artificial Analysis ↗

AA-Omniscience v1.1 · hallucination rate ↗64.07%

DeepSeek V4.1 Flash 1017 (Max) · AA-Omniscience v1.1 · hallucination rate, lower is better ↗

1 more reported result

66.75% · DeepSeek V4.1 Flash 1017 (Max) · AA-Omniscience · hallucination rate, lower is better · Artificial Analysis ↗

GDPval-AA v2.1 ↗1745.77 Elo

DeepSeek V4.1 Flash 1017 (Max) · GDPval-AA v2.1 · Elo ↗

1 more reported result

1746 Elo · DeepSeek-V4.1-Flash-1017 (max) · GDPval-AA v2.1 · GDPval-AA v2.1 (Elo-score) · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

FrontierCode 1.1 ↗43.66%

DeepSeek V4.1 Flash 1017 (Max) · FrontierCode 1.1 · AA implementation ↗

AA-LCR v1.1 ↗82.67%

DeepSeek V4.1 Flash 1017 (Max) · AA-LCR v1.1 ↗

SciCode ↗55.90%

DeepSeek V4.1 Flash 1017 (Max) · SciCode · AA implementation ↗

Humanity’s Last Exam ↗43.33%

DeepSeek V4.1 Flash 1017 (Max) · Humanity’s Last Exam · AA implementation ↗

3 more reported results

42.6 (44.0†) · DS-V4.1-Flash-1017 · HLE w/o tools · DeepSeek V4.1 Flash 1017 model card ↗

58.8% · DeepSeek-V4.1-Flash-1017 (max) · Humanity’s Last Exam · HLE-Full w/ tools · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

42.6% · DeepSeek-V4.1-Flash-1017 (max) · Humanity’s Last Exam · HLE-Full · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

MMMU-Pro ↗80.64%

DeepSeek V4.1 Flash 1017 (Max) · MMMU-Pro · AA implementation ↗

2 more reported results

82.5% · DeepSeek-V4.1-Flash-1017 (max) · MMMU-Pro · MMMU-Pro w/ python · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

80.6% · DeepSeek-V4.1-Flash-1017 (max) · MMMU-Pro · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

MLCR-AA ↗26.67%

DeepSeek V4.1 Flash 1017 (Max) · MLCR-AA · overall ↗

CritPt v2 ↗50.00%

DeepSeek V4.1 Flash 1017 (Max) · CritPt v2 · code tools ↗

CritPt ↗24.00%

DeepSeek V4.1 Flash 1017 (Max) · CritPt · original ↗

AA-PersonalityBench ↗44.15 percentile

DeepSeek V4.1 Flash 1017 (Max) · AA-PersonalityBench · expressiveness ↗

7 more reported results

84.77 percentile · DeepSeek V4.1 Flash 1017 (Max) · AA-PersonalityBench · organization · Artificial Analysis ↗

67.00 percentile · DeepSeek V4.1 Flash 1017 (Max) · AA-PersonalityBench · informativeness · Artificial Analysis ↗

87.31 percentile · DeepSeek V4.1 Flash 1017 (Max) · AA-PersonalityBench · transparency · Artificial Analysis ↗

18.77 percentile · DeepSeek V4.1 Flash 1017 (Max) · AA-PersonalityBench · warmth · Artificial Analysis ↗

94.92 percentile · DeepSeek V4.1 Flash 1017 (Max) · AA-PersonalityBench · conviction · Artificial Analysis ↗

31.46 percentile · DeepSeek V4.1 Flash 1017 (Max) · AA-PersonalityBench · imagination · Artificial Analysis ↗

41.62 percentile · DeepSeek V4.1 Flash 1017 (Max) · AA-PersonalityBench · slop · Artificial Analysis ↗

DeepSWE v1.1 ↗77.0

DS-V4.1-Flash-1017 · DeepSWE v1.1 ↗

1 more reported result

77% · DeepSeek-V4.1-Flash-1017 (max) · DeepSWE v1.1 · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

ProgramBench ↗29.1

DS-V4.1-Flash-1017 · ProgramBench ↗

1 more reported result

29.1% · DeepSeek-V4.1-Flash-1017 (max) · ProgramBench · Program Bench · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

NL2Repo-Bench ↗68.2

DS-V4.1-Flash-1017 · NL2Repo-Bench ↗

AutomationBench · Zapier ↗60.5

DS-V4.1-Flash-1017 · Automation-Bench v1.0.6 ↗

Agents’ Last Exam ↗35.4

DS-V4.1-Flash-1017 · Agents’ Last Exam ↗

2 more reported results

37.7% · DeepSeek-V4.1-Flash-1017 (max) · Agents’ Last Exam · Agents' Last Exam (ALL) · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

36% · DeepSeek-V4.1-Flash-1017 (max) · Agents’ Last Exam · Agents' Last Exam (CLI) · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

Apex Bench v1.1 ↗72.1

DS-V4.1-Flash-1017 · Apex Bench v1.1 ↗

Chartography ↗83.0

DS-V4.1-Flash-1017 · Chartography (w/tools) ↗

1 more reported result

83% · DeepSeek-V4.1-Flash-1017 (max) · Chartography · ChartoGraphy w/ python · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

BabyVision ↗92.5

DS-V4.1-Flash-1017 · BabyVision (w/tools) ↗

1 more reported result

92.5% · DeepSeek-V4.1-Flash-1017 (max) · BabyVision · BabyVision w/ python · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

ZeroBench ↗51.0

DS-V4.1-Flash-1017 · ZeroBench-main (w/tools) ↗

3 more reported results

51% · DeepSeek-V4.1-Flash-1017 (max) · ZeroBench · ZeroBench_main w/ python (pass@5) · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

11.8% · DeepSeek-V4.1-Flash-1017 (max) · ZeroBench · ZeroBench_main · avg@5 · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

23% · DeepSeek-V4.1-Flash-1017 (max) · ZeroBench · ZeroBench_main · pass@5 · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

ErdősBench ↗3.040 / 4

DeepSeek V4.1 Flash 1017 · ErdősBench · mean progress / 4 · API ↗

GPQA Diamond ↗93.18%

DeepSeek V4.1 Flash 1017 (max) · GPQA diamond · Best score (across scorers) · % ↗

1 more reported result

92.2% · DeepSeek-V4.1-Flash-1017 (max) · GPQA Diamond · GPQA-Diamond · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

FrontierMath Tiers 1-3 (v2) ↗71.23%

DeepSeek V4.1 Flash 1017 (max) · FrontierMath-Tiers-1-3-v2-Private · Best score (across scorers) · % ↗

FrontierMath Tier 4 (v2) ↗46.34%

DeepSeek V4.1 Flash 1017 (max) · FrontierMath-Tier-4-v2-Private · Best score (across scorers) · % ↗

OTIS Mock AIME 2024-2025 ↗99.44%

DeepSeek V4.1 Flash 1017 (max) · OTIS Mock AIME 2024-2025 · Best score (across scorers) · % ↗

SimpleQA Verified ↗42.5%

DeepSeek V4.1 Flash 1017 (max) · SimpleQA Verified · Best score (across scorers) · % ↗

Chess Puzzles ↗46%

DeepSeek V4.1 Flash 1017 (max) · Chess Puzzles · Best score (across scorers) · % ↗

Mystery Game Puzzles ↗49%

DeepSeek V4.1 Flash 1017 (max) · Mystery Game Puzzles · Best score (across scorers) · % ↗

Vals Index ↗61.26 %

DeepSeek V4.1 Flash 1017 · Vals Index · Accuracy ↗

Vals Finance Index ↗60.89 %

DeepSeek V4.1 Flash 1017 · Vals Finance Index · Accuracy ↗

Vals Legal Index ↗71.86 %

DeepSeek V4.1 Flash 1017 · Vals Legal Index · Accuracy ↗

Legal Research Bench ↗45.19 %

DeepSeek V4.1 Flash 1017 · Legal Research Bench · All-Pass ↗

HLAB-Vals-1.1 ↗14.17 %

DeepSeek V4.1 Flash 1017 · HLAB-Vals-1.1 · Task Pass Rate ↗

Harvey Legal Agent Benchmark ↗9.17 %

DeepSeek V4.1 Flash 1017 · Harvey's Legal Agent Benchmark v1 · Accuracy ↗

1 more reported result

21% · DeepSeek-V4.1-Flash-1017 (max) · Harvey Legal Agent Benchmark · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

Vals Finance Agent v2.1 ↗62.93 %

DeepSeek V4.1 Flash 1017 · Finance Agent · Accuracy ↗

Excel Modeling Benchmark (EMB & EMB v1.1) ↗69.17 %

DeepSeek V4.1 Flash 1017 · Excel Modeling Benchmark · Accuracy ↗

1 more reported result

61.8 % · DeepSeek V4.1 Flash 1017 · Excel Modeling Benchmark v1 · Accuracy · Vals AI · Excel Modeling Benchmark v1 ↗

Tax Agent Bench ↗44.35 %

DeepSeek V4.1 Flash 1017 · Tax Agent Bench · Accuracy ↗

Vals Finance Agent v2 ↗56.96 %

DeepSeek V4.1 Flash 1017 · Finance Agent v2 · Accuracy ↗

1 more reported result

54.5% · DeepSeek-V4.1-Flash-1017 (max) · Vals Finance Agent v2 · Frontier Finance · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

Vibe Code Bench ↗87.73 %

DeepSeek V4.1 Flash 1017 (OpenHands) · Vibe Code Bench · original task set ↗

Code Migration v1.1 ↗67.52 %

DeepSeek V4.1 Flash 1017 · Code Migration · Accuracy ↗

1 more reported result

44.21 % · DeepSeek V4.1 Flash 1017 · Code Migration v1 · Accuracy · Vals AI · Code Migration v1 ↗

SWE-Marathon ↗40%

DeepSeek-V4.1-Flash-1017 (max) · SWE-Marathon · SWE Marathon v1.1 · Kimi K3.1 Pro · complete release comparison ↗

BenchCAD ↗41.2%

DeepSeek-V4.1-Flash-1017 (max) · BenchCAD · BenchCAD w/o tools · Kimi K3.1 Pro · complete release comparison ↗

1 more reported result

81.6% · DeepSeek-V4.1-Flash-1017 (max) · BenchCAD · BenchCAD w/ tools · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

PostTrainBench ↗30.5%

DeepSeek-V4.1-Flash-1017 (max) · PostTrainBench · PostTrain Bench v1.2 · Kimi K3.1 Pro · complete release comparison ↗

Kimi Code Bench 2.0 ↗76.1%

DeepSeek-V4.1-Flash-1017 (max) · Kimi Code Bench 2.0 · Kimi Code Bench 2.0 (Internal) · Kimi K3.1 Pro · complete release comparison ↗

Kimi Code Bench 3.0 ↗51.5%

DeepSeek-V4.1-Flash-1017 (max) · Kimi Code Bench 3.0 · Kimi Code Bench 3.0 Long-horizon (Internal) · Kimi K3.1 Pro · complete release comparison ↗

BioMysteryBench ↗80%

DeepSeek-V4.1-Flash-1017 (max) · BioMysteryBench · Kimi K3.1 Pro · complete release comparison ↗

CharXiv ↗85.6%

DeepSeek-V4.1-Flash-1017 (max) · CharXiv · CharXiv (RQ) w/ python · Kimi K3.1 Pro · complete release comparison ↗

1 more reported result

77% · DeepSeek-V4.1-Flash-1017 (max) · CharXiv · CharXiv (RQ) · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

APEX-Agents · original ↗63.4%

DeepSeek-V4.1-Flash-1017 (max) · APEX-Agents · original · Kimi K3.1 Pro · complete release comparison ↗

OfficeQA Pro ↗40.4%

DeepSeek-V4.1-Flash-1017 (max) · OfficeQA Pro · OfficeQA Pro v2 ↗

1 more reported result

62.4% · DeepSeek-V4.1-Flash-1017 (max) · OfficeQA Pro · OfficeQA Pro v1 · Introducing Kimi K3.1 Pro ↗

HANDBOOK.md ↗28.9%

DeepSeek-V4.1-Flash-1017 (max) · HANDBOOK.md · Kimi K3.1 Pro · complete release comparison ↗

OSWorld 2.1 ↗69.3%

DeepSeek-V4.1-Flash-1017 (max) · OSWorld 2.1 · OSWorld V2.1 · Kimi K3.1 Pro · complete release comparison ↗

DECK-Arena ↗1231 Elo

DeepSeek-V4.1-Flash-1017 (max) · DECK-Arena · DECK-Arena (Internal, Elo-score) · Kimi K3.1 Pro · complete release comparison ↗

Design Bench ↗863 Elo

DeepSeek-V4.1-Flash-1017 (max) · Design Bench · Design Bench (Internal, Elo-score) · Kimi K3.1 Pro · complete release comparison ↗

Kimi Hallucination Benchmark ↗23.3%

DeepSeek-V4.1-Flash-1017 (max) · Kimi Hallucination Benchmark · Kimi Hallu Bench Hard w/ tools (Internal, lower is better) · Kimi K3.1 Pro · complete release comparison ↗

1 more reported result

13.1% · DeepSeek-V4.1-Flash-1017 (max) · Kimi Hallucination Benchmark · Kimi Hallu Bench w/o tools (Internal, lower is better) · Kimi K3.1 Pro · complete release comparison · Introducing Kimi K3.1 Pro ↗

GDPx v1.0 ↗5.4%

DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Overall · intended grading ↗

53 more reported results

3.7% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Eight-domain subset · intended grading · GDPx v1.0 — SemiAnalysis ↗

9.0% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Recursive Self-Improvement · intended grading · GDPx v1.0 — SemiAnalysis ↗

0.2% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Chip Design · intended grading · GDPx v1.0 — SemiAnalysis ↗

16.7% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Cybersecurity · intended grading · GDPx v1.0 — SemiAnalysis ↗

12.8% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Biology · intended grading · GDPx v1.0 — SemiAnalysis ↗

0.0% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Finance · intended grading · GDPx v1.0 — SemiAnalysis ↗

0.0% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Legal · intended grading · GDPx v1.0 — SemiAnalysis ↗

15.4% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Quant Trading · intended grading · GDPx v1.0 — SemiAnalysis ↗

2.5% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Hardware Design · intended grading · GDPx v1.0 — SemiAnalysis ↗

2.7% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Materials Science · intended grading · GDPx v1.0 — SemiAnalysis ↗

0.0% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Alignment · intended grading · GDPx v1.0 — SemiAnalysis ↗

0.0% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Software Engineering · intended grading · GDPx v1.0 — SemiAnalysis ↗

0.0% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Aria · Chip Design · intended grading · GDPx v1.0 — SemiAnalysis ↗

0.0% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Arcverex · Chip Design · intended grading · GDPx v1.0 — SemiAnalysis ↗

0.7% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Ergodex · Chip Design · intended grading · GDPx v1.0 — SemiAnalysis ↗

16.7% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Arithmetic · Cybersecurity · intended grading · GDPx v1.0 — SemiAnalysis ↗

12.8% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · LatchBio · Biology · intended grading · GDPx v1.0 — SemiAnalysis ↗

0.0% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Living Context · Finance · intended grading · GDPx v1.0 — SemiAnalysis ↗

0.0% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Chakra · Legal · intended grading · GDPx v1.0 — SemiAnalysis ↗

15.4% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Third Moment · Quant Trading · intended grading · GDPx v1.0 — SemiAnalysis ↗

9.0% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Extrapolate · Recursive Self-Improvement · intended grading · GDPx v1.0 — SemiAnalysis ↗

2.5% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Beautiful Machines · Hardware Design · intended grading · GDPx v1.0 — SemiAnalysis ↗

1.3% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · ValenceForge · Materials Science · intended grading · GDPx v1.0 — SemiAnalysis ↗

4.2% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Emulated · Materials Science · intended grading · GDPx v1.0 — SemiAnalysis ↗

0.0% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Fleet · Alignment · intended grading · GDPx v1.0 — SemiAnalysis ↗

0.0% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Datacurve · Software Engineering · intended grading · GDPx v1.0 — SemiAnalysis ↗

10.6% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Overall · partial credit · GDPx v1.0 — SemiAnalysis ↗

9.7% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Eight-domain subset · partial credit · GDPx v1.0 — SemiAnalysis ↗

9.0% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Recursive Self-Improvement · partial credit · GDPx v1.0 — SemiAnalysis ↗

0.4% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Chip Design · partial credit · GDPx v1.0 — SemiAnalysis ↗

16.7% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Cybersecurity · partial credit · GDPx v1.0 — SemiAnalysis ↗

22.2% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Biology · partial credit · GDPx v1.0 — SemiAnalysis ↗

26.2% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Finance · partial credit · GDPx v1.0 — SemiAnalysis ↗

18.7% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Legal · partial credit · GDPx v1.0 — SemiAnalysis ↗

15.4% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Quant Trading · partial credit · GDPx v1.0 — SemiAnalysis ↗

2.5% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Hardware Design · partial credit · GDPx v1.0 — SemiAnalysis ↗

5.5% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Materials Science · partial credit · GDPx v1.0 — SemiAnalysis ↗

0.0% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Alignment · partial credit · GDPx v1.0 — SemiAnalysis ↗

0.0% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Software Engineering · partial credit · GDPx v1.0 — SemiAnalysis ↗

0.5% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Aria · Chip Design · partial credit · GDPx v1.0 — SemiAnalysis ↗

0.0% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Arcverex · Chip Design · partial credit · GDPx v1.0 — SemiAnalysis ↗

0.7% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Ergodex · Chip Design · partial credit · GDPx v1.0 — SemiAnalysis ↗

16.7% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Arithmetic · Cybersecurity · partial credit · GDPx v1.0 — SemiAnalysis ↗

22.2% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · LatchBio · Biology · partial credit · GDPx v1.0 — SemiAnalysis ↗

26.2% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Living Context · Finance · partial credit · GDPx v1.0 — SemiAnalysis ↗

18.7% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Chakra · Legal · partial credit · GDPx v1.0 — SemiAnalysis ↗

15.4% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Third Moment · Quant Trading · partial credit · GDPx v1.0 — SemiAnalysis ↗

9.0% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Extrapolate · Recursive Self-Improvement · partial credit · GDPx v1.0 — SemiAnalysis ↗

2.5% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Beautiful Machines · Hardware Design · partial credit · GDPx v1.0 — SemiAnalysis ↗

4.2% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · ValenceForge · Materials Science · partial credit · GDPx v1.0 — SemiAnalysis ↗

6.8% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Emulated · Materials Science · partial credit · GDPx v1.0 — SemiAnalysis ↗

0.0% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Fleet · Alignment · partial credit · GDPx v1.0 — SemiAnalysis ↗

0.0% · DeepSeek V4.1 Flash 1017 · GDPx v1.0 · Datacurve · Software Engineering · partial credit · GDPx v1.0 — SemiAnalysis ↗

Closest models by score

Similar positions on shared benchmark leaderboards. Smaller gaps are closer.

Qwen3.8 Max (0902) →Qwen

17.5 percentile-point gap

25 of 40 shared benchmark comparisons

GLM-5.3 →Z.ai

17.8 percentile-point gap

25 of 40 shared benchmark comparisons

Claude Haiku 5.5 →Anthropic

18.2 percentile-point gap

38 of 40 shared benchmark comparisons

How similarity is measured

Mean absolute gap in percentile positions on shared benchmark settings; lower is closer. At least 60% overlap is required, with a minimum of three when available. Independent evaluations are used when there are at least three; otherwise published report comparisons are used. Repeated model observations within one setting are averaged before ranking. Writing-style traits and composite indices are excluded. This describes similarity, not an overall quality ranking.

About the model report

The model card labels tool-enabled visual evaluations and names specific releases. Preserve these conditions when reading its comparison table. Reviewed 2026-10-25.

Read the official model report ↗

11 benchmarks in the report

Search this collection →

COD / BENCHMARK115 reported results

Terminal-Bench 4.0

Complex terminal work spans coding, science, infrastructure, and other technical tasks, with executable checks.

Verified task success↗

RSN / BENCHMARK131 reported results

Humanity’s Last Exam

Expert-authored questions stretch academic reasoning and knowledge across many disciplines.

Answer accuracy↗

COD / BENCHMARK97 reported results

DeepSWE v1.1

Original engineering problems test coding agents on substantial changes in existing repositories.

Verified task success↗

COD / BENCHMARK81 reported results

ProgramBench

Agents reconstruct programs from observed behavior rather than receiving a conventional repository repair issue.

Behavioral reconstruction score↗

AGT / BENCHMARK24 reported results

Agents’ Last Exam

Professional software workflows test whether an agent can deliver the requested result in a real application environment.

Professional task success↗

COD / BENCHMARK6 reported results

NL2Repo-Bench

A natural-language software request is used to assess whether an agent can produce a working repository.

Repository creation score↗

AGT / BENCHMARK6 reported results

Apex Bench v1.1

This named professional-task evaluation appears in DeepSeek’s recent model comparison.

Source-specific task score↗

RSN / BENCHMARK17 reported results

Chartography

Specialist charts test visual reasoning grounded in professional practice.

Professional chart accuracy↗

RSN / BENCHMARK20 reported results

BabyVision

Visual problem solving tests whether a model can reason over image structure rather than rely only on familiar textual knowledge.

Visual reasoning accuracy↗

RSN / BENCHMARK26 reported results

ZeroBench

Challenging visual questions probe failures that remain after ordinary image understanding becomes strong.

Visual task accuracy↗

AGT / BENCHMARK20 reported results

AutomationBench · Zapier

Business workflows require an agent to coordinate real tools and leave the simulated business systems in the required final state.

Verified final-state success↗

COMPARISON CHECKLIST

Match benchmark version, task subset, model effort, harness, tool access, attempt budget, scoring rule, and fallback behavior. A shared benchmark name is the start of a comparison, not proof that two reported scores use the same contract.

DeepSeek original report ↗