Kimi K2 0905: benchmark results
September 5, 2025 update of Moonshot's open Kimi K2 MoE with improved agentic coding and a 256K context. Provider: Moonshot. Released 2025-09-05. Access: Open.
Aggregate Rating 1662 ± 6, rank #288 of 1656 rated models, from 555 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AGC-Bench - future_ideas | 2.07 | Dataset z-score | 100 |
| AGC-Bench - hypogen | 2.19 | Dataset z-score | 100 |
| AGC-Bench - pron_vs_prompt | 1.41 | Dataset z-score | 100 |
| AGC-Bench - thenextchapter | 2.77 | Dataset z-score | 100 |
| DSGym - DAEval-Verified | 92.82 | Exact-Match Accuracy (%) | 100 |
| DSGym - DSBio | 43.33 | Exact-Match Accuracy (%) | 100 |
| DSGym - QRData-Verified | 63.68 | Exact-Match Accuracy (%) | 100 |
| Mask of Civility (Combined) | 88 | Accuracy (%) of the combined (the definitions and the few-sh | 100 |
| OpenCompass LLM - Knowledge - Chinese (CompassBench 2510) | 89 | Score (%) | 100 |
| AGC-Bench - conceptual_design | 2.02 | Dataset z-score | 98.8 |
| AGC-Bench - cpers | 1.4 | Dataset z-score | 98.8 |
| AGC-Bench - cue_word_story | 1.16 | Dataset z-score | 98.8 |
Interactive version: theaggregate.ai/model?slug=kimi-k2-0905 · How It Works · Data refreshed daily, snapshot 2026-10-20.