KAT-Coder-Pro V2

Composite indexes

• Artificial Analysis Intelligence Index#72 of 151
  22independentv4.3.2 (AA estimate; independent evaluation forthcoming), non-reasoningArtificial Analysis: KAT Coder Pro V2 ↗

  Leader: Claude Opus 5.5 at 58

Coding

• Terminal-Bench 2.x#36 of 105
  70%independentAA, Terminal-Bench 2.1, non-reasoningArtificial Analysis: KAT Coder Pro V2 ↗

  Leader: Claude Fable 5.1 at 91.4%

Agents and tool use

• τ²-Bench#22 of 89
  89.5%independentAA, telecom, non-reasoningArtificial Analysis: KAT Coder Pro V2 ↗

  Leader: GLM-5.2 at 99.1%

• GDPval-AA#74 of 109
  720independentGDPval-AA v2.1, non-reasoningArtificial Analysis: KAT Coder Pro V2 ↗

  Leader: Muse Wonder Full at 1857

Reasoning and knowledge

• Humanity’s Last Exam#84 of 153
  16.1%independentAA, no tools, non-reasoningArtificial Analysis: KAT Coder Pro V2 ↗

  Leader: Claude Sonnet 5.5 at 64.5%

• GPQA Diamond#58 of 144
  85.5%independentAA, non-reasoningArtificial Analysis: KAT Coder Pro V2 ↗

  Leader: GPT-6 Astra at 96.1%

• CritPt#107 of 144
  0%independentAA, non-reasoningArtificial Analysis: KAT Coder Pro V2 ↗

  Leader: Muse Wonder Full at 33%

Vision and long context

• AA Long Context Reasoning#72 of 144
  73%independentAA, non-reasoningArtificial Analysis: KAT Coder Pro V2 ↗

  Leader: Kimi K3 at 88.7%

Other published results