KAT Coder Pro V2 — benchmark results

Kwaipilot's flagship agentic-coding KAT-Coder successor (MoE, 72B active). Provider: Kuaishou. Released 2026-03-27. Access: API.

Unified ELO 1689 ± 28, rank #295 of 1776 rated models, from 41 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AA Terminal-Bench Hard49.24Accuracy (%)94.5
AA-LCR66Score (self-reported)87
AA GPQA Diamond85.45Accuracy (%)84.7
AA Long Context Reasoning66Accuracy (%)84.1
AA TAU-2 Bench89.47Accuracy (%)83.9
Artificial Analysis Intelligence Index33.69Intelligence Index81.2
AA IFBench66.67Accuracy (%)79.5
Wolfram LLM Benchmarking Project53.8Correct Functionality (%)79.5
SnakeBench27TrueSkill Rating77
AA Humanity's Last Exam15.99Accuracy (%)75
AA Omniscience-21.6Score70.1
AA Omniscience - Software Engineering (SWE) - C49Accuracy (%)68.8

Interactive version: theaggregate.ai/model?slug=kat-coder-pro-v2 · How the rankings work · Data refreshed daily, snapshot 2026-07-22.