RLTasksby ulam.ai
Models / DeepSeek

DeepSeek / MODEL EVIDENCE

DeepSeek V4 Flash 0731

41 benchmarks · 47 observations · 42 sources · released 2026-07-31

Scores at a glance

41 benchmarks · large dot: DeepSeek V4 Flash 0731; small dots: closest models where reported

Chess Puzzles ↗32%
076

DeepSeek V4 Flash 0731 (max) · Chess Puzzles · Best score (across scorers) · % ↗

CritPt ↗16.57%
032.29

DeepSeek V4 Flash 0731 (max) · CritPt · Accuracy · % ↗

LMCA ↗41.73%
2.7868.26

DeepSeek V4 Flash 0731 (max) · LMCA · Score · % ↗

DTBench ↗90.93%
41.6598.93

DeepSeek V4 Flash 0731 (max) · DTBench · Accuracy · % ↗

SciCode ↗49.88%
1.2766.9

DeepSeek V4 Flash 0731 (max) · SciCode · Score · % ↗

MedCode ↗41.41 %
19.7263.57

DeepSeek V4 Flash 0731 · MedCode · Accuracy ↗

MedScribe ↗80.36 %
4.2791.43

DeepSeek V4 Flash 0731 · MedScribe · Accuracy ↗

MMLU-Pro ↗86.21 %
30.2892.38

DeepSeek V4 Flash 0731 · MMLU Pro · Accuracy ↗

Vibe Code Bench ↗74.74 %
092.39

DeepSeek V4 Flash 0731 (OpenHands) · Vibe Code Bench · original task set ↗

SkillsBench ↗50.67 %
18.0669.8

DeepSeek V4 Flash 0731 (OpenHands) · SkillsBench · Score ↗

ALE-Bench ↗1306.08 Performance points
137.782951.32

deepseek-v4-flash-0731-max · Average contest performance ↗

Closest models by score

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

14.6 percentile-point gap

24 of 31 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.

Filter comparisons to this model →

Sources

Every plotted observation links to its evaluation. Reported configurations, benchmark versions and task subsets can differ; similarity describes shared scores, not training lift.

  1. ARC Prize - Leaderboard ↗

    ARC Prize Foundation

  2. Arena leaderboard — WebDev evaluation ↗

    Arena

  3. SciCode Benchmark | Artificial Analysis ↗

    Artificial Analysis

  4. 𝜏³-Banking Benchmark | Artificial Analysis ↗

    Artificial Analysis

  5. DTBench — Conceptual Reasoning Index ↗

    Conceptual Reasoning Index

  6. LMCA — Conceptual Reasoning Index ↗

    Conceptual Reasoning Index

  7. Ale Bench — methodology and published results ↗

    Epoch AI · 2026

  8. ARC-AGI-2 ↗

    Epoch AI

  9. Chess Puzzles ↗

    Epoch AI

  10. CritPt v1 ↗

    Epoch AI

  11. FrontierMath Tier 4 (v2) ↗

    Epoch AI

  12. FrontierMath Tiers 1-3 (v2) ↗

    Epoch AI

  13. GPQA Diamond ↗

    Epoch AI

  14. Mystery Game Puzzles ↗

    Epoch AI

  15. OTIS Mock AIME 2024-2025 ↗

    Epoch AI

  16. SimpleQA Verified ↗

    Epoch AI

  17. AI benchmarks & capabilities ↗

    Epoch AI

  18. WeirdML ↗

    Håvard Tveit Ihle

  19. BioMysteryBench — evaluation methodology and results ↗

    Vals AI · 2026

  20. Code Migration v1 — evaluation methodology and results ↗

    Vals AI · 2026

  21. CorpFin — evaluation methodology and results ↗

    Vals AI · 2026

  22. CyberBench — evaluation methodology and results ↗

    Vals AI · 2026

  23. Excel Modeling Benchmark v1 — evaluation methodology and results ↗

    Vals AI · 2026

  24. Finance Agent v2 — evaluation methodology and results ↗

    Vals AI · 2026

  25. GPQA Diamond — evaluation methodology and results ↗

    Vals AI · 2026

  26. Harvey's Legal Agent Benchmark v1 — evaluation methodology and results ↗

    Vals AI · 2026

  27. IOI — evaluation methodology and results ↗

    Vals AI · 2026

  28. LiveCodeBench — evaluation methodology and results ↗

    Vals AI · 2026

  29. LegalBench — evaluation methodology and results ↗

    Vals AI · 2026

  30. Legal Research Bench — evaluation methodology and results ↗

    Vals AI · 2026

  31. MedCode — evaluation methodology and results ↗

    Vals AI · 2026

  32. MedScribe — evaluation methodology and results ↗

    Vals AI · 2026

  33. MMLU Pro — evaluation methodology and results ↗

    Vals AI · 2026

  34. ProgramBench — evaluation methodology and results ↗

    Vals AI · 2026

  35. ProofBench — evaluation methodology and results ↗

    Vals AI · 2026

  36. SkillsBench — evaluation methodology and results ↗

    Vals AI · 2026

  37. SWE-bench Verified — evaluation methodology and results ↗

    Vals AI · 2026

  38. Tax Agent Bench — evaluation methodology and results ↗

    Vals AI · 2026

  39. TaxEval — evaluation methodology and results ↗

    Vals AI · 2026

  40. Terminal-Bench 4.0 — evaluation methodology and results ↗

    Vals AI · 2026

  41. Terminal-Bench-Science — evaluation methodology and results ↗

    Vals AI · 2026

  42. Vibe Code Bench — evaluation methodology and results ↗

    Vals AI · 2026

Names used in source tables

DeepSeek V4 Flash 0731 · DeepSeek V4 Flash 0731 (Max) · DeepSeek V4 Flash 0731 (Mini-SWE-agent) · DeepSeek V4 Flash 0731 (OpenHands) · DeepSeek V4 Flash 0731 (high) · DeepSeek V4 Flash 0731 (low) · DeepSeek V4 Flash 0731 (max) · deepseek-v4-flash-0731-max