GeekRadar / The comparison desk
COMPARE.
AI models
deepseek

DeepSeek-V3.2-Exp

AI model Released

Context window
163,840 tokens
Open weights
Yes

Capability & preference

General capability · ECI

A cross-benchmark capability estimate. Index points, not a percentage; intervals show uncertainty.

DeepSeek-V3.2-Exp 145.0* DeepSeek-V3.2-Exp · 95% interval 142.5–147.1

Human preference · Arena

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

DeepSeek-V3.2-Exp 1422.2 DeepSeek-V3.2-Exp · 95% interval 1415.8–1428.5 · 11,996 votes · 2026-10-28

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 · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,461.57 rating

Confidence interval: 1,443.60 – 1,479.54

1031 votes

Category
chinese
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / codinghuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,435.04 rating

Confidence interval: 1,423.52 – 1,446.57

2549 votes

Category
coding
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / creative_writinghuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,403.21 rating

Confidence interval: 1,388.76 – 1,417.67

1664 votes

Category
creative_writing
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / englishhuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,440.35 rating

Confidence interval: 1,431.68 – 1,449.01

5043 votes

Category
english
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / exclude_tieshuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,413.80 rating

Confidence interval: 1,405.05 – 1,422.55

8416 votes

Category
exclude_ties
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / experthuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,403.49 rating

Confidence interval: 1,381.44 – 1,425.55

664 votes

Category
expert
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / frenchhuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,433.78 rating

Confidence interval: 1,394.34 – 1,473.23

227 votes

Category
french
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / germanhuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,439.68 rating

Confidence interval: 1,402.89 – 1,476.47

227 votes

Category
german
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / hard_promptshuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,428.90 rating

Confidence interval: 1,421.01 – 1,436.79

6582 votes

Category
hard_prompts
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / hard_prompts_englishhuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,436.94 rating

Confidence interval: 1,425.91 – 1,447.97

2766 votes

Category
hard_prompts_english
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / industry_business_and_management_and_financial_operationshuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,412.22 rating

Confidence interval: 1,399.78 – 1,424.65

2249 votes

Category
industry_business_and_management_and_financial_operations
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / industry_entertainment_and_sports_and_mediahuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,407.10 rating

Confidence interval: 1,394.71 – 1,419.48

2185 votes

Category
industry_entertainment_and_sports_and_media
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / industry_legal_and_governmenthuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,406.48 rating

Confidence interval: 1,386.18 – 1,426.78

796 votes

Category
industry_legal_and_government
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / industry_life_and_physical_and_social_sciencehuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,430.16 rating

Confidence interval: 1,416.87 – 1,443.45

1937 votes

Category
industry_life_and_physical_and_social_science
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / industry_mathematicalhuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,415.01 rating

Confidence interval: 1,392.69 – 1,437.33

636 votes

Category
industry_mathematical
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / industry_medicine_and_healthcarehuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,427.37 rating

Confidence interval: 1,403.91 – 1,450.82

596 votes

Category
industry_medicine_and_healthcare
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / industry_software_and_it_serviceshuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,435.88 rating

Confidence interval: 1,426.45 – 1,445.31

4145 votes

Category
industry_software_and_it_services
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / industry_writing_and_literature_and_languagehuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,407.32 rating

Confidence interval: 1,395.93 – 1,418.70

2662 votes

Category
industry_writing_and_literature_and_language
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / instruction_followinghuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,400.69 rating

Confidence interval: 1,390.49 – 1,410.89

3392 votes

Category
instruction_following
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / japanesehuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,331.49 rating

Confidence interval: 1,282.00 – 1,380.97

148 votes

Category
japanese
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / koreanhuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,370.23 rating

Confidence interval: 1,330.66 – 1,409.80

212 votes

Category
korean
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / longer_queryhuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,417.49 rating

Confidence interval: 1,406.86 – 1,428.13

3061 votes

Category
longer_query
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / mathhuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,418.28 rating

Confidence interval: 1,398.31 – 1,438.26

793 votes

Category
math
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / multi_turnhuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,421.41 rating

Confidence interval: 1,408.49 – 1,434.34

1992 votes

Category
multi_turn
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / non_englishhuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,404.16 rating

Confidence interval: 1,396.35 – 1,411.96

6919 votes

Category
non_english
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / overallhuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,422.16 rating

Confidence interval: 1,415.82 – 1,428.49

11996 votes

Category
overall
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / polishhuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,420.15 rating

Confidence interval: 1,392.63 – 1,447.66

440 votes

Category
polish
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / russianhuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,403.16 rating

Confidence interval: 1,386.70 – 1,419.62

1174 votes

Category
russian
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / spanishhuman preference · 2026-10-28 · rating

DeepSeek-V3.2-Exp

1,441.18 rating

Confidence interval: 1,410.32 – 1,472.04

355 votes

Category
spanish
Evaluated configuration
DeepSeek-V3.2-Exp
Evaluation type
human preference
Arena text preference / chinesehuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,461.34 rating

Confidence interval: 1,443.36 – 1,479.31

1030 votes

Category
chinese
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / codinghuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,435.22 rating

Confidence interval: 1,423.70 – 1,446.73

2552 votes

Category
coding
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / creative_writinghuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,403.73 rating

Confidence interval: 1,389.28 – 1,418.17

1665 votes

Category
creative_writing
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / englishhuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,440.23 rating

Confidence interval: 1,431.56 – 1,448.89

5047 votes

Category
english
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / exclude_tieshuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,413.85 rating

Confidence interval: 1,405.10 – 1,422.60

8420 votes

Category
exclude_ties
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / experthuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,403.50 rating

Confidence interval: 1,381.48 – 1,425.53

665 votes

Category
expert
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / frenchhuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,433.28 rating

Confidence interval: 1,393.81 – 1,472.75

227 votes

Category
french
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / germanhuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,439.64 rating

Confidence interval: 1,402.84 – 1,476.44

227 votes

Category
german
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / hard_promptshuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,428.85 rating

Confidence interval: 1,420.97 – 1,436.74

6584 votes

Category
hard_prompts
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / hard_prompts_englishhuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,436.90 rating

Confidence interval: 1,425.87 – 1,447.93

2766 votes

Category
hard_prompts_english
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / industry_business_and_management_and_financial_operationshuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,412.10 rating

Confidence interval: 1,399.67 – 1,424.54

2250 votes

Category
industry_business_and_management_and_financial_operations
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / industry_entertainment_and_sports_and_mediahuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,407.26 rating

Confidence interval: 1,394.87 – 1,419.65

2184 votes

Category
industry_entertainment_and_sports_and_media
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / industry_legal_and_governmenthuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,406.39 rating

Confidence interval: 1,386.10 – 1,426.67

798 votes

Category
industry_legal_and_government
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / industry_life_and_physical_and_social_sciencehuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,430.24 rating

Confidence interval: 1,416.95 – 1,443.54

1939 votes

Category
industry_life_and_physical_and_social_science
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / industry_mathematicalhuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,414.40 rating

Confidence interval: 1,392.07 – 1,436.73

635 votes

Category
industry_mathematical
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / industry_medicine_and_healthcarehuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,427.42 rating

Confidence interval: 1,403.96 – 1,450.88

596 votes

Category
industry_medicine_and_healthcare
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / industry_software_and_it_serviceshuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,436.00 rating

Confidence interval: 1,426.58 – 1,445.42

4148 votes

Category
industry_software_and_it_services
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / industry_writing_and_literature_and_languagehuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,407.52 rating

Confidence interval: 1,396.14 – 1,418.90

2663 votes

Category
industry_writing_and_literature_and_language
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / instruction_followinghuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,400.70 rating

Confidence interval: 1,390.50 – 1,410.90

3391 votes

Category
instruction_following
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / japanesehuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,331.25 rating

Confidence interval: 1,281.75 – 1,380.74

148 votes

Category
japanese
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / koreanhuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,372.08 rating

Confidence interval: 1,332.66 – 1,411.49

212 votes

Category
korean
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / longer_queryhuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,417.62 rating

Confidence interval: 1,406.99 – 1,428.25

3064 votes

Category
longer_query
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / mathhuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,418.18 rating

Confidence interval: 1,398.21 – 1,438.15

793 votes

Category
math
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / multi_turnhuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,421.49 rating

Confidence interval: 1,408.56 – 1,434.43

1992 votes

Category
multi_turn
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / non_englishhuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,404.27 rating

Confidence interval: 1,396.46 – 1,412.07

6923 votes

Category
non_english
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / overallhuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,422.18 rating

Confidence interval: 1,415.84 – 1,428.51

12004 votes

Category
overall
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / polishhuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,420.09 rating

Confidence interval: 1,392.57 – 1,447.61

440 votes

Category
polish
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / russianhuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,403.62 rating

Confidence interval: 1,387.15 – 1,420.09

1173 votes

Category
russian
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference
Arena text preference / spanishhuman preference · 2026-10-15 · rating

DeepSeek-V3.2-Exp

1,440.07 rating

Confidence interval: 1,409.18 – 1,470.96

354 votes

Category
spanish
Evaluated configuration
deepseek-v3.2-exp
Evaluation type
human preference

Run it locally

Inference · memory estimate

DeepSeek-V3.2-Exp

Not enough verified architecture information for a reliable VRAM estimate.

Capabilities

Context window
163,840 tokens
Open weights
Yes
Model license
mit

Input modalities

  • text

Output modalities

  • text

Processing & platform

Architecture
deepseek_v32

* Provisional data