Granite 4.2 3B

Composite indexes

• Artificial Analysis Intelligence Index#104 of 142
  9independentIntelligence Index v4.3; reasoningArtificial Analysis: Granite 4.2 3B ↗

  Leader: Claude Opus 5.5 at 58

Coding

• SciCode#89 of 94
  25.3%independentreasoningArtificial Analysis: Granite 4.2 3B ↗
  24.1%vendorthinkingHugging Face: ibm-granite/granite-4.2-30b model card ↗

  Leader: Claude Opus 5.5 at 66.9%

• Terminal-Bench 4.0#85 of 88
  0%independentreasoningArtificial Analysis: Granite 4.2 3B ↗

  Leader: Claude Mythos 5.1 at 60.9%

• Terminal-Bench 2.x#88 of 103
  13.9%independentTerminal-Bench 2.1; reasoningArtificial Analysis: Granite 4.2 3B ↗

  Leader: Claude Fable 5.1 at 91.4%

• LiveCodeBench#24 of 56
  69.7%vendorv6; thinkingHugging Face: ibm-granite/granite-4.2-30b model card ↗

  Leader: Fugu at 92.9%

Agents and tool use

• GDPval-AA#94 of 103
  188independentreasoningArtificial Analysis: Granite 4.2 3B ↗

  Leader: Claude Opus 5.5 at 1846

Reasoning and knowledge

• GPQA Diamond#118 of 142
  55.9%independentreasoningArtificial Analysis: Granite 4.2 3B ↗
  54.8%vendorGPQA; thinkingHugging Face: ibm-granite/granite-4.2-30b model card ↗

  Leader: GPT-6 Astra at 96.1%

• Humanity’s Last Exam#112 of 147
  6.6%independentno tools; reasoningArtificial Analysis: Granite 4.2 3B ↗

  Leader: Claude Opus 5.5 at 61.4%

• CritPt#123 of 139
  0%independentreasoningArtificial Analysis: Granite 4.2 3B ↗

  Leader: GPT-5.6 Sol at 32.3%

• HMMT February 2026#8 of 8
  66.7%vendorHMMT Feb 2025 (not 2026); thinkingHugging Face: ibm-granite/granite-4.2-30b model card ↗

  Leader: GLM-5.2 at 92.5%

• MMLU-Pro#13 of 14
  67.8%vendorthinkingHugging Face: ibm-granite/granite-4.2-30b model card ↗

  Leader: Nemotron 3 Ultra 550B A55B at 86.8%

Vision and long context

• AA Long Context Reasoning#110 of 139
  24.3%independentreasoningArtificial Analysis: Granite 4.2 3B ↗

  Leader: Kimi K3 at 88.7%

Other published results