RLTasksby ulam.ai
Models / Google

Google / MODEL EVIDENCE

Gemini 3 Flash Preview

29 benchmarks · 45 observations · 31 sources · released 2025-12-17

Scores at a glance

29 benchmarks · large dot: Gemini 3 Flash Preview; small dots: closest models where reported

GPQA Diamond ↗89.39%
89.3989.39

Gemini 3 Flash Preview · GPQA diamond · Best score (across scorers) · % ↗

1 more reported result

83.21% · Gemini 3 Flash Preview · GPQA diamond · Best score (across scorers) · % · Epoch AI · GPQA diamond ↗

Chess Puzzles ↗39%
076

Gemini 3 Flash Preview · Chess Puzzles · Best score (across scorers) · % ↗

BALROG ↗48.1%
3.768.3

Gemini 3 Flash Preview · Balrog · Average progress · % ↗

GeoBench ↗88%
7588

Gemini 3 Flash Preview · GeoBench · ACW Country % · % ↗

GSO ↗9.8%
9.888.24

Gemini 3 Flash Preview · GSO-Bench · Score OPT@1 · % ↗

SimpleBench ↗61.1%
45.880.6

Gemini 3 Flash Preview · SimpleBench · Score (AVG@5) · % ↗

VPCT ↗72.6%
3091

Gemini 3 Flash Preview · VPCT · Correct · % ↗

LMCA ↗43.13%
2.7868.26

Gemini 3 Flash Preview · LMCA · Score · % ↗

DTBench ↗89.07%
41.6598.93

Gemini 3 Flash Preview · DTBench · Accuracy · % ↗

GDP.pdf ↗10%
1010

Gemini 3 Flash Preview · GDP.pdf · GDP.pdf score · % ↗

EconBench ↗1.8296 error points
1.831.83

gemini-3-flash-preview · EconBench · EconBench prototype · risk-consistency error · published prototype observations ↗

ForecastBench ↗64.7 Brier index
46.467

gemini-3-flash-preview† · Overall difficulty-adjusted Brier index ↗

Closest models by score

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

GPT-5 →OpenAI
14.0 percentile-point gap

7 of 8 shared benchmark comparisons

19.8 percentile-point gap

8 of 8 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. Blueprint-Bench 2 | Andon Labs ↗

    Andon Labs

  2. ARC Prize - Leaderboard ↗

    ARC Prize Foundation

  3. Arena leaderboard — WebDev evaluation ↗

    Arena

  4. BALROG: Benchmarking Agentic LLM and VLM Reasoning On Games ↗

    Paglieri, Davide, Cupiał, Bartłomiej, Coward, Samuel, Piterbarg, Ulyana, Wolczyk, Maciej, Khan, Akbir et al. · 2024

    All 13 authors

    Paglieri, Davide, Cupiał, Bartłomiej, Coward, Samuel, Piterbarg, Ulyana, Wolczyk, Maciej, Khan, Akbir, Pignatelli, Eduardo, Kuciński, Łukasz, Pinto, Lerrel, Fergus, Rob, Foerster, Jakob Nicolaus, Parker-Holder, Jack, Rocktäschel, Tim

  5. DTBench — Conceptual Reasoning Index ↗

    Conceptual Reasoning Index

  6. LMCA — Conceptual Reasoning Index ↗

    Conceptual Reasoning Index

  7. EconBench: Economic choice by language models — gemini-3-flash-preview prototype metrics ↗

    EconBench project · 2026

  8. ARC-AGI-1 ↗

    Epoch AI

  9. ARC-AGI-2 ↗

    Epoch AI

  10. Chess Puzzles ↗

    Epoch AI

  11. Forecastbench — methodology and published results ↗

    Epoch AI · 2026

  12. FrontierMath Tier 4 (v1, old) ↗

    Epoch AI

  13. FrontierMath Tier 4 (v2) ↗

    Epoch AI

  14. FrontierMath Tiers 1-3 (v1, old) ↗

    Epoch AI

  15. FrontierMath Tiers 1-3 (v2) ↗

    Epoch AI

  16. GeoBench ↗

    Epoch AI

  17. GPQA Diamond ↗

    Epoch AI

  18. Mystery Game Puzzles ↗

    Epoch AI

  19. OTIS Mock AIME 2024-2025 ↗

    Epoch AI

  20. SimpleQA Verified ↗

    Epoch AI

  21. SWE-bench Verified ↗

    Epoch AI

  22. Terminal-Bench 2.0 ↗

    Epoch AI

  23. Capabilities & benchmarking — retained external evaluation records ↗

    Epoch AI · 2026

  24. FutureSearch Evals ↗

    FutureSearch

  25. GSO: Challenging Software Optimization Tasks for Evaluating SWE-Agents ↗

    Shetty, Manish, Jain, Naman, Liu, Jinjian, Kethanaboyina, Vijay, Sen, Koushik, Stoica, Ion · 2025

  26. WeirdML ↗

    Håvard Tveit Ihle

  27. Scale Labs — evaluation methodology and results ↗

    Scale Labs / SEAL · 2026

  28. AI Model Benchmarks — LM Council ↗

    LM Council

  29. GDP.pdf Benchmark | Surge AI ↗

    Surge AI

  30. Terminal-Bench 2.0 leaderboard ↗

    Terminal-Bench Team

  31. ProofBench — evaluation methodology and results ↗

    Vals AI · 2026

Names used in source tables

Gemini 3 Flash Preview · gemini-3-flash-preview · gemini-3-flash-preview†