GeekRadar / The comparison desk
COMPARE.
AI models
moonshot

kimi-k2-0905-preview

Model evaluation

Capability & preference

Human preference · Arena

Overall text preferences from head-to-head votes. Not factual accuracy or a task-success rate.

kimi-k2-0905-preview 1380.0 kimi-k2-0905-preview · 95% interval 1373.5–1386.5 · 11,589 votes · 2026-10-15

Recorded results

Only the same benchmark, version, workload, scenario and unit share a row. Missing results are not zero. System results include the full submitted hardware and software configuration.

Arena text preference / chinesehuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,415.40 rating

Confidence interval: 1,396.26 – 1,434.55

957 votes

Category
chinese
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / codinghuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,399.02 rating

Confidence interval: 1,386.62 – 1,411.43

2233 votes

Category
coding
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / creative_writinghuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,349.57 rating

Confidence interval: 1,334.35 – 1,364.80

1482 votes

Category
creative_writing
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / englishhuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,378.81 rating

Confidence interval: 1,370.01 – 1,387.62

4778 votes

Category
english
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / exclude_tieshuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,351.81 rating

Confidence interval: 1,343.32 – 1,360.31

8431 votes

Category
exclude_ties
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / experthuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,365.31 rating

Confidence interval: 1,341.32 – 1,389.30

576 votes

Category
expert
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / frenchhuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,380.50 rating

Confidence interval: 1,341.85 – 1,419.14

216 votes

Category
french
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / germanhuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,386.68 rating

Confidence interval: 1,351.76 – 1,421.61

278 votes

Category
german
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / hard_promptshuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,383.63 rating

Confidence interval: 1,375.29 – 1,391.97

5554 votes

Category
hard_prompts
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / hard_prompts_englishhuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,378.18 rating

Confidence interval: 1,366.11 – 1,390.26

2391 votes

Category
hard_prompts_english
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / industry_business_and_management_and_financial_operationshuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,353.33 rating

Confidence interval: 1,340.20 – 1,366.47

2047 votes

Category
industry_business_and_management_and_financial_operations
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / industry_entertainment_and_sports_and_mediahuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,342.03 rating

Confidence interval: 1,329.22 – 1,354.84

2077 votes

Category
industry_entertainment_and_sports_and_media
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / industry_legal_and_governmenthuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,402.84 rating

Confidence interval: 1,381.83 – 1,423.85

769 votes

Category
industry_legal_and_government
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / industry_life_and_physical_and_social_sciencehuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,387.06 rating

Confidence interval: 1,372.80 – 1,401.31

1739 votes

Category
industry_life_and_physical_and_social_science
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / industry_mathematicalhuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,399.49 rating

Confidence interval: 1,376.89 – 1,422.09

641 votes

Category
industry_mathematical
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / industry_medicine_and_healthcarehuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,382.75 rating

Confidence interval: 1,360.21 – 1,405.28

719 votes

Category
industry_medicine_and_healthcare
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / industry_software_and_it_serviceshuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,393.82 rating

Confidence interval: 1,384.12 – 1,403.53

3862 votes

Category
industry_software_and_it_services
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / industry_writing_and_literature_and_languagehuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,345.70 rating

Confidence interval: 1,334.04 – 1,357.36

2576 votes

Category
industry_writing_and_literature_and_language
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / instruction_followinghuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,347.96 rating

Confidence interval: 1,336.99 – 1,358.94

2888 votes

Category
instruction_following
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / japanesehuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,348.52 rating

Confidence interval: 1,308.53 – 1,388.51

209 votes

Category
japanese
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / koreanhuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,326.14 rating

Confidence interval: 1,291.44 – 1,360.84

312 votes

Category
korean
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / longer_queryhuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,353.23 rating

Confidence interval: 1,341.76 – 1,364.70

2574 votes

Category
longer_query
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / mathhuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,397.31 rating

Confidence interval: 1,376.67 – 1,417.96

755 votes

Category
math
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / multi_turnhuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,365.31 rating

Confidence interval: 1,352.22 – 1,378.41

2029 votes

Category
multi_turn
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / non_englishhuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,372.55 rating

Confidence interval: 1,364.79 – 1,380.30

6811 votes

Category
non_english
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / overallhuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,380.00 rating

Confidence interval: 1,373.51 – 1,386.49

11589 votes

Category
overall
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / polishhuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,377.15 rating

Confidence interval: 1,354.50 – 1,399.80

679 votes

Category
polish
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / russianhuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,384.30 rating

Confidence interval: 1,364.34 – 1,404.25

839 votes

Category
russian
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference
Arena text preference / spanishhuman preference · 2026-10-15 · rating

kimi-k2-0905-preview

1,386.34 rating

Confidence interval: 1,355.08 – 1,417.61

391 votes

Category
spanish
Evaluated configuration
kimi-k2-0905-preview
Evaluation type
human preference

Run it locally

Inference · memory estimate

kimi-k2-0905-preview

Not enough verified architecture information for a reliable VRAM estimate.

Capabilities

Model license
Modified MIT