Ministral 3 14B

Composite indexes

• Artificial Analysis Intelligence Index#129 of 142
  6independentIntelligence Index v4.3; non-reasoningArtificial Analysis: Ministral 3 14B ↗

  Leader: Claude Opus 5.5 at 58

Coding

• SciCode#90 of 94
  23.8%independentnon-reasoningArtificial Analysis: Ministral 3 14B ↗

  Leader: Claude Opus 5.5 at 66.9%

• Terminal-Bench 4.0#78 of 88
  0%independentnon-reasoningArtificial Analysis: Ministral 3 14B ↗

  Leader: Claude Mythos 5.1 at 60.9%

• Terminal-Bench 2.x#92 of 103
  9.7%independentTerminal-Bench 2.1; non-reasoningArtificial Analysis: Ministral 3 14B ↗

  Leader: Claude Fable 5.1 at 91.4%

• LiveCodeBench#40 of 56
  35.1%independentnon-reasoningArtificial Analysis: Ministral 3 14B ↗

  Leader: Fugu at 92.9%

Agents and tool use

• τ²-Bench#57 of 88
  27.2%independenttelecom; non-reasoningArtificial Analysis: Ministral 3 14B ↗

  Leader: GLM-5.2 at 99.1%

• GDPval-AA#91 of 103
  234independentnon-reasoningArtificial Analysis: Ministral 3 14B ↗

  Leader: Claude Opus 5.5 at 1846

Reasoning and knowledge

• GPQA Diamond#116 of 142
  57.2%independentnon-reasoningArtificial Analysis: Ministral 3 14B ↗

  Leader: GPT-6 Astra at 96.1%

• Humanity’s Last Exam#128 of 147
  4.6%independentno tools; non-reasoningArtificial Analysis: Ministral 3 14B ↗

  Leader: Claude Opus 5.5 at 61.4%

• CritPt#106 of 139
  0%independentnon-reasoningArtificial Analysis: Ministral 3 14B ↗

  Leader: GPT-5.6 Sol at 32.3%

Vision and long context

• AA Long Context Reasoning#108 of 139
  26.3%independentnon-reasoningArtificial Analysis: Ministral 3 14B ↗

  Leader: Kimi K3 at 88.7%

• MMMU-Pro#65 of 72
  49.8%independentnon-reasoningArtificial Analysis: Ministral 3 14B ↗

  Leader: Claude Opus 5.5 at 87.7%

Other published results