Kimi Linear 48B A3B Instruct — benchmark results

Moonshot's 48B MoE (3B active) with hybrid Kimi Delta linear attention and a 1M context, MIT-licensed. Provider: Moonshot. Released 2025-10-30. Access: Open.

Unified ELO 1439 ± 56, rank #1048 of 1776 rated models, from 22 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
UGI - Writing37.74Writing Score65.5
UGI - Natural Intelligence22.09NatInt Score51.8
AA Terminal-Bench Hard11.36Accuracy (%)46.8
AA-LCR25.7Score (self-reported)45.6
AA LiveCodeBench37.78Pass@1 (%)44.7
AA Long Context Reasoning25.67Accuracy (%)37.7
AA AIME 202536.33Accuracy (%)36.5
SpeechMap Compliance53.4% Requests Completed36.1
Artificial Analysis Intelligence Index8.53Intelligence Index27.7
ATLAS33.1ATLAScore (self-reported)22.7
AA MMLU-Pro58.5Accuracy (%)21.8
SvelteBench66.7Average pass@1 (%)21.6

Interactive version: theaggregate.ai/model?slug=kimi-linear-48b-a3b-instruct · How the rankings work · Data refreshed daily, snapshot 2026-07-22.