INTELLECT-3

Composite indexes

• Artificial Analysis Intelligence Index#97 of 142
  11independentIntelligence Index v4.3; reasoning; estimatedArtificial Analysis: INTELLECT-3 (based on GLM-4.5-Air) ↗

  Leader: Claude Opus 5.5 at 58

Coding

• LiveCodeBench#12 of 56
  77.7%independentreasoningArtificial Analysis: INTELLECT-3 (based on GLM-4.5-Air) ↗

  Leader: Fugu at 92.9%

Agents and tool use

• τ²-Bench#59 of 88
  26.6%independenttelecom; reasoningArtificial Analysis: INTELLECT-3 (based on GLM-4.5-Air) ↗

  Leader: GLM-5.2 at 99.1%

Reasoning and knowledge

• GPQA Diamond#83 of 142
  76.1%independentreasoningArtificial Analysis: INTELLECT-3 (based on GLM-4.5-Air) ↗

  Leader: GPT-6 Astra at 96.1%

• Humanity’s Last Exam#85 of 147
  13.1%independentno tools; reasoningArtificial Analysis: INTELLECT-3 (based on GLM-4.5-Air) ↗

  Leader: Claude Opus 5.5 at 61.4%

• CritPt#88 of 139
  0.3%independentreasoningArtificial Analysis: INTELLECT-3 (based on GLM-4.5-Air) ↗

  Leader: GPT-5.6 Sol at 32.3%

Vision and long context

• AA Long Context Reasoning#96 of 139
  39.7%independentreasoningArtificial Analysis: INTELLECT-3 (based on GLM-4.5-Air) ↗

  Leader: Kimi K3 at 88.7%

Other published results