Gemma 4 26B A4B

Composite indexes

• Artificial Analysis Intelligence Index#74 of 142
  17independentv4.3.2, reasoningArtificial Analysis: Gemma 4 26B A4B (Reasoning) ↗
  13independentv4.3.2, non-reasoningArtificial Analysis: Gemma 4 26B A4B (Non-reasoning) ↗

  Leader: Claude Opus 5.5 at 58

Coding

• SciCode#70 of 94
  40%independentAA run, reasoningArtificial Analysis: Gemma 4 26B A4B (Reasoning) ↗

  Leader: Claude Opus 5.5 at 66.9%

• Terminal-Bench 2.x#72 of 103
  39%independentAA run (Terminal-Bench 2.1), reasoningArtificial Analysis: Gemma 4 26B A4B (Reasoning) ↗

  Leader: Claude Fable 5.1 at 91.4%

• LiveCodeBench#13 of 56
  77.1%vendorLiveCodeBench v6Gemma 4 model card ↗

  Leader: Fugu at 92.9%

Agents and tool use

• GDPval-AA#74 of 103
  553independentGDPval-AA v2.1, reasoningArtificial Analysis: Gemma 4 26B A4B (Reasoning) ↗

  Leader: Claude Opus 5.5 at 1846

• τ²-Bench#49 of 88
  68.2%vendoraverage over 3 domainsGemma 4 model card ↗
  43.6%independentAA run, telecom, reasoningArtificial Analysis: Gemma 4 26B A4B (Reasoning) ↗

  Leader: GLM-5.2 at 99.1%

Reasoning and knowledge

• GPQA Diamond#73 of 142
  82.3%vendorGemma 4 model card ↗
  79.2%independentAA run, reasoningArtificial Analysis: Gemma 4 26B A4B (Reasoning) ↗

  Leader: GPT-6 Astra at 96.1%

• Humanity’s Last Exam#74 of 147
  19.3%independentAA run, no tools, reasoningArtificial Analysis: Gemma 4 26B A4B (Reasoning) ↗
  8.7%vendorno toolsGemma 4 model card ↗

  Leader: Claude Opus 5.5 at 61.4%

• CritPt#95 of 139
  0%independentAA run, reasoningArtificial Analysis: Gemma 4 26B A4B (Reasoning) ↗

  Leader: GPT-5.6 Sol at 32.3%

• MMLU-Pro#5 of 14
  82.6%vendorinstruction-tunedGemma 4 model card ↗

  Leader: Nemotron 3 Ultra 550B A55B at 86.8%

• AIME 2026#9 of 13
  88.3%vendorno toolsGemma 4 model card ↗

  Leader: ERNIE 5.1 at 99.6%

Vision and long context

• AA Long Context Reasoning#77 of 139
  65.7%independentAA-LCR v1.1, reasoningArtificial Analysis: Gemma 4 26B A4B (Reasoning) ↗

  Leader: Kimi K3 at 88.7%

• MMMU-Pro#45 of 72
  73.8%vendorGemma 4 model card ↗
  69.2%independentAA run, reasoningArtificial Analysis: Gemma 4 26B A4B (Reasoning) ↗

  Leader: Claude Opus 5.5 at 87.7%

Human preference

• LMArena Text#39 of 54
  1438independentgemma-4-26b-a4b; 5,804 votesLMArena text leaderboard ↗

  Leader: Claude Fable 5 at 1506

• LMArena WebDev#49 of 54
  1362independentgemma-4-26b-a4b; 1,441 votes (preliminary)LMArena Code Arena | WebDev leaderboard ↗

  Leader: Claude Opus 5.5 at 1818

Other published results