RLTasksby ulam.ai
Models / Google

Google / MODEL EVIDENCE

Gemini 3 Pro Preview

41 benchmarks · 51 observations · 44 sources · released 2025-11-18

Scores at a glance

41 benchmarks · large dot: Gemini 3 Pro Preview; small dots: closest models where reported

MMLU-Pro ↗89.76%
84.8689.76

Gemini 3 Pro Preview (High) · MMLU-Pro · AA implementation ↗

LiveCodeBench ↗91.75%
79.7991.75

Gemini 3 Pro Preview (High) · LiveCodeBench · AA implementation ↗

GPQA Diamond ↗92.61%
9.2896.4

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

BALROG ↗58.1%
3.768.3

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

GeoBench ↗84%
3286

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

GSO ↗18.63%
026.5

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

Terminal-Bench 2.0 ↗69.43%
3.0784.72

Gemini 3 Pro Preview · Terminal Bench · Accuracy mean · % ↗

7 more reported results

65.17% · Gemini 3 Pro Preview · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗

62.25% · Gemini 3 Pro Preview · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗

61.8% · Gemini 3 Pro Preview · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗

61.12% · Gemini 3 Pro Preview · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗

56.85% · Gemini 3 Pro Preview · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗

55.96% · Gemini 3 Pro Preview · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗

62.2% · Gemini 3 Pro Preview · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗

VPCT ↗91%
3091

Gemini 3 Pro Preview · VPCT · Correct · % ↗

CritPt ↗6.9%
032.29

Gemini 3 Pro Preview · CritPt · Accuracy · % ↗

GDPval ↗40.3%
9.949.7

Gemini 3 Pro Preview · GDPval · Win Rate (%) · % ↗

CL-bench ↗15.8%
11.427.9

Gemini 3 Pro Preview · CL-bench · Overall · % ↗

Fortress ↗41.69%
5.7874.39

gemini-3-pro-preview · Scale Labs · Risk elicitation score ↗

VISTA ↗51.49%
15.1854.65

gemini-3-pro-preview · Scale Labs · Score ↗

HLE-Rolling ↗51.4022%
51.451.4

Gemini 3 Pro Preview · HLE-Rolling · FUSE 649-item subset · Mean response correctness · FUSE HLE-Verifications · Gemini 3 Pro Preview · 50 responses/item ↗

1 more reported result

100% · Gemini 3 Pro Preview · HLE-Rolling · FUSE 649-item subset · At least one correct response in 50 · FUSE HLE-Verifications · Gemini 3 Pro Preview · 50 responses/item · HLE with Gemini 3 Pro — FUSE HLE-Verifications ↗

AlgoTune ↗1.83 ×
1.312.05

Gemini 3 Pro Preview (high) · Harmonic-mean speedup ↗

ForecastBench ↗65.3 Brier index
46.467

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

Closest models by score

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

There are not enough shared scored settings to identify close comparisons reliably.

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. AIME 2025 Benchmark | Artificial Analysis ↗

    Artificial Analysis

  4. Global-MMLU-Lite Benchmark | Artificial Analysis ↗

    Artificial Analysis

  5. LiveCodeBench Benchmark | Artificial Analysis ↗

    Artificial Analysis

  6. MMLU-Pro Benchmark | Artificial Analysis ↗

    Artificial Analysis

  7. OfficeQA Pro: An Enterprise Benchmark for End-to-End Grounded Reasoning — published result table ↗

    Opsahl-Ong, Krista, Singhvi, Arnav, Collins, Jasmine, Zhou, Ivan, Wang, Cindy, Baheti, Ashutosh et al. · 2026

    All 13 authors

    Opsahl-Ong, Krista, Singhvi, Arnav, Collins, Jasmine, Zhou, Ivan, Wang, Cindy, Baheti, Ashutosh, Oertell, Owen, Portes, Jacob, Havens, Sam, Elsen, Erich, Bendersky, Michael, Zaharia, Matei, Chen, Xing

  8. 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

  9. Algotune — methodology and published results ↗

    Epoch AI · 2026

  10. ARC-AGI-2 ↗

    Epoch AI

  11. CL-bench ↗

    Epoch AI

  12. CritPt v1 ↗

    Epoch AI

  13. Forecastbench — methodology and published results ↗

    Epoch AI · 2026

  14. FrontierMath Tier 4 (v1, old) ↗

    Epoch AI

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

    Epoch AI

  16. GDPval ↗

    Epoch AI

  17. GPQA Diamond ↗

    Epoch AI

  18. Humanity's Last Exam ↗

    Epoch AI

  19. METR Time Horizons ↗

    Epoch AI

  20. OTIS Mock AIME 2024-2025 ↗

    Epoch AI

  21. Remote Labor Index ↗

    Epoch AI

  22. SWE-bench Verified ↗

    Epoch AI

  23. Terminal-Bench 2.0 ↗

    Epoch AI

  24. Capabilities & benchmarking — retained external evaluation records ↗

    Epoch AI · 2026

  25. FutureSearch Evals ↗

    FutureSearch

  26. GeoBench — GeoGuessr model benchmark ↗

    GeoBench

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

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

  28. WeirdML ↗

    Håvard Tveit Ihle

  29. HLE with Gemini 3 Pro — FUSE HLE-Verifications ↗

    FUSE-verifiers · 2026

  30. Scale Labs — evaluation methodology and results ↗

    Scale Labs / SEAL · 2026

  31. EnigmaEval ↗

    Scale Labs / SEAL

  32. Scale Labs — evaluation methodology and results ↗

    Scale Labs / SEAL · 2026

  33. Scale Labs — evaluation methodology and results ↗

    Scale Labs / SEAL · 2026

  34. Scale Labs — evaluation methodology and results ↗

    Scale Labs / SEAL · 2026

  35. Scale Labs — evaluation methodology and results ↗

    Scale Labs / SEAL · 2026

  36. Scale Labs — evaluation methodology and results ↗

    Scale Labs / SEAL · 2026

  37. Scale Labs — evaluation methodology and results ↗

    Scale Labs / SEAL · 2026

  38. Scale Labs — evaluation methodology and results ↗

    Scale Labs / SEAL · 2026

  39. Scale Labs — evaluation methodology and results ↗

    Scale Labs / SEAL · 2026

  40. Scale Labs — evaluation methodology and results ↗

    Scale Labs / SEAL · 2026

  41. AI Model Benchmarks — LM Council ↗

    LM Council

  42. Metr Autonomy Task Horizon Suite — methodology and published results ↗

    METR · 2026

  43. Terminal-Bench 2.0 leaderboard ↗

    Terminal-Bench Team

  44. ProofBench — evaluation methodology and results ↗

    Vals AI · 2026

Names used in source tables

Gemini 3 Pro Preview · Gemini 3 Pro Preview (High) · Gemini 3 Pro Preview (high) · Gemini 3 Pro Preview (low) · gemini-3-pro-preview · gemini-3-pro-preview (Thinking) · gemini-3-pro-preview†