Kimi K2 Instruct (0905): benchmark results

Provider: Moonshot. Released 2025-09-05. Access: Open.

Unified ELO 1744 ± 22, rank #252 of 2656 rated models, from 35 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
TuRTLe - Icarus Syntax94.22Average Score (%)95.5
TuRTLe - Verilator Syntax95.09Average Score (%)93.2
BALSAM - Text Classification40.31Overall score (0-100, LLM-judged generation and multiple cho85.7
Wolfram LLM Benchmarking Project57.2Correct Functionality (%)84.4
TuRTLe - Verilator Performance67.03Average Score (%)84.1
TuRTLe - Verilator Synthesis68.68Average Score (%)84.1
BALSAM - Creative Writing51.19Overall score (0-100, LLM-judged generation and multiple cho82.1
TuRTLe - Icarus Performance67.47Average Score (%)81.8
TuRTLe - Icarus Synthesis69.25Average Score (%)81.8
TuRTLe Code Completion (Icarus Verilog)71.77Aggregated Score (self-reported)81.4
TuRTLe Code Completion (Verilator)71.79Aggregated Score (self-reported)81.4
TuRTLe Spec-to-RTL (Icarus Verilog)68.72Aggregated Score (self-reported)81.4

Interactive version: theaggregate.ai/model?slug=kimi-k2-instruct-0905 · How It Works · Data refreshed daily, snapshot 2026-09-19.