Gemma 4 31B

Composite indexes

• Artificial Analysis Intelligence Index#68 of 142
  19independentv4.3.2, reasoningArtificial Analysis: Gemma 4 31B (Reasoning) ↗
  14independentv4.3.2, non-reasoningArtificial Analysis: Gemma 4 31B (Non-reasoning) ↗

  Leader: Claude Opus 5.5 at 58

Coding

• SciCode#54 of 94
  45.5%independentAA run, reasoningArtificial Analysis: Gemma 4 31B (Reasoning) ↗

  Leader: Claude Opus 5.5 at 66.9%

• Terminal-Bench 4.0#66 of 88
  0%independentAA run (Terminal-Bench 4.0), reasoningArtificial Analysis: Gemma 4 31B (Reasoning) ↗

  Leader: Claude Mythos 5.1 at 60.9%

• Terminal-Bench 2.x#66 of 103
  43.4%independentAA run (Terminal-Bench 2.1), reasoningArtificial Analysis: Gemma 4 31B (Reasoning) ↗

  Leader: Claude Fable 5.1 at 91.4%

• LiveCodeBench#10 of 56
  80%vendorLiveCodeBench v6Gemma 4 model card ↗

  Leader: Fugu at 92.9%

Agents and tool use

• GDPval-AA#71 of 103
  606independentGDPval-AA v2.1, reasoningArtificial Analysis: Gemma 4 31B (Reasoning) ↗

  Leader: Claude Opus 5.5 at 1846

• τ²-Bench#41 of 88
  76.9%vendoraverage over 3 domainsGemma 4 model card ↗
  59.9%independentAA run, telecom, reasoningArtificial Analysis: Gemma 4 31B (Reasoning) ↗

  Leader: GLM-5.2 at 99.1%

Reasoning and knowledge

• GPQA Diamond#55 of 142
  85.7%independentAA run, reasoningArtificial Analysis: Gemma 4 31B (Reasoning) ↗
  84.3%vendorGemma 4 model card ↗

  Leader: GPT-6 Astra at 96.1%

• Humanity’s Last Exam#64 of 147
  23.6%independentAA run, no tools, reasoningArtificial Analysis: Gemma 4 31B (Reasoning) ↗
  19.5%vendorno toolsGemma 4 model card ↗

  Leader: Claude Opus 5.5 at 61.4%

• CritPt#67 of 139
  1.4%independentAA run, reasoningArtificial Analysis: Gemma 4 31B (Reasoning) ↗

  Leader: GPT-5.6 Sol at 32.3%

• MMLU-Pro#2 of 14
  85.2%vendorinstruction-tunedGemma 4 model card ↗

  Leader: Nemotron 3 Ultra 550B A55B at 86.8%

• AIME 2026#8 of 13
  89.2%vendorno toolsGemma 4 model card ↗

  Leader: ERNIE 5.1 at 99.6%

Vision and long context

• AA Long Context Reasoning#75 of 139
  69.7%independentAA-LCR v1.1, reasoningArtificial Analysis: Gemma 4 31B (Reasoning) ↗

  Leader: Kimi K3 at 88.7%

• MMMU-Pro#42 of 72
  76.9%vendorGemma 4 model card ↗
  73.4%independentAA run, reasoningArtificial Analysis: Gemma 4 31B (Reasoning) ↗

  Leader: Claude Opus 5.5 at 87.7%

Human preference

• LMArena Text#34 of 54
  1451independentgemma-4-31b; 5,894 votesLMArena text leaderboard ↗

  Leader: Claude Fable 5 at 1506

• LMArena WebDev#48 of 54
  1364independentgemma-4-31b; 11,636 votesLMArena Code Arena | WebDev leaderboard ↗

  Leader: Claude Opus 5.5 at 1818

Other published results