GeekRadar / The comparison desk
COMPARE.
AI models
Not reported

RWKV-4 Raven 14B

Model evaluation

Capability & preference

Human preference · Arena

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

RWKV-4 Raven 14B 948.4 RWKV-4 Raven 14B · 95% interval 937.1–959.6 · 4,845 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

RWKV-4 Raven 14B

910.52 rating

Confidence interval: 872.55 – 948.49

218 votes

Category
chinese
Evaluated configuration
RWKV-4 Raven 14B
Evaluation type
human preference
Arena text preference / codinghuman preference · 2026-10-28 · rating

RWKV-4 Raven 14B

927.57 rating

Confidence interval: 901.02 – 954.11

505 votes

Category
coding
Evaluated configuration
RWKV-4 Raven 14B
Evaluation type
human preference
Arena text preference / creative_writinghuman preference · 2026-10-28 · rating

RWKV-4 Raven 14B

933.46 rating

Confidence interval: 911.26 – 955.67

874 votes

Category
creative_writing
Evaluated configuration
RWKV-4 Raven 14B
Evaluation type
human preference
Arena text preference / englishhuman preference · 2026-10-28 · rating

RWKV-4 Raven 14B

967.98 rating

Confidence interval: 955.49 – 980.48

3997 votes

Category
english
Evaluated configuration
RWKV-4 Raven 14B
Evaluation type
human preference
Arena text preference / exclude_tieshuman preference · 2026-10-28 · rating

RWKV-4 Raven 14B

720.99 rating

Confidence interval: 703.97 – 738.02

3301 votes

Category
exclude_ties
Evaluated configuration
RWKV-4 Raven 14B
Evaluation type
human preference
Arena text preference / hard_promptshuman preference · 2026-10-28 · rating

RWKV-4 Raven 14B

895.96 rating

Confidence interval: 876.03 – 915.89

876 votes

Category
hard_prompts
Evaluated configuration
RWKV-4 Raven 14B
Evaluation type
human preference
Arena text preference / hard_prompts_englishhuman preference · 2026-10-28 · rating

RWKV-4 Raven 14B

901.28 rating

Confidence interval: 879.53 – 923.03

771 votes

Category
hard_prompts_english
Evaluated configuration
RWKV-4 Raven 14B
Evaluation type
human preference
Arena text preference / industry_business_and_management_and_financial_operationshuman preference · 2026-10-28 · rating

RWKV-4 Raven 14B

888.28 rating

Confidence interval: 854.52 – 922.03

360 votes

Category
industry_business_and_management_and_financial_operations
Evaluated configuration
RWKV-4 Raven 14B
Evaluation type
human preference
Arena text preference / industry_entertainment_and_sports_and_mediahuman preference · 2026-10-28 · rating

RWKV-4 Raven 14B

902.59 rating

Confidence interval: 880.98 – 924.19

834 votes

Category
industry_entertainment_and_sports_and_media
Evaluated configuration
RWKV-4 Raven 14B
Evaluation type
human preference
Arena text preference / industry_legal_and_governmenthuman preference · 2026-10-28 · rating

RWKV-4 Raven 14B

975.86 rating

Confidence interval: 928.15 – 1,023.58

190 votes

Category
industry_legal_and_government
Evaluated configuration
RWKV-4 Raven 14B
Evaluation type
human preference
Arena text preference / industry_life_and_physical_and_social_sciencehuman preference · 2026-10-28 · rating

RWKV-4 Raven 14B

927.71 rating

Confidence interval: 904.40 – 951.03

844 votes

Category
industry_life_and_physical_and_social_science
Evaluated configuration
RWKV-4 Raven 14B
Evaluation type
human preference
Arena text preference / industry_mathematicalhuman preference · 2026-10-28 · rating

RWKV-4 Raven 14B

918.97 rating

Confidence interval: 893.18 – 944.76

488 votes

Category
industry_mathematical
Evaluated configuration
RWKV-4 Raven 14B
Evaluation type
human preference
Arena text preference / industry_medicine_and_healthcarehuman preference · 2026-10-28 · rating

RWKV-4 Raven 14B

944.42 rating

Confidence interval: 896.50 – 992.33

177 votes

Category
industry_medicine_and_healthcare
Evaluated configuration
RWKV-4 Raven 14B
Evaluation type
human preference
Arena text preference / industry_software_and_it_serviceshuman preference · 2026-10-28 · rating

RWKV-4 Raven 14B

914.57 rating

Confidence interval: 895.00 – 934.13

1072 votes

Category
industry_software_and_it_services
Evaluated configuration
RWKV-4 Raven 14B
Evaluation type
human preference
Arena text preference / industry_writing_and_literature_and_languagehuman preference · 2026-10-28 · rating

RWKV-4 Raven 14B

933.94 rating

Confidence interval: 914.90 – 952.97

1098 votes

Category
industry_writing_and_literature_and_language
Evaluated configuration
RWKV-4 Raven 14B
Evaluation type
human preference
Arena text preference / instruction_followinghuman preference · 2026-10-28 · rating

RWKV-4 Raven 14B

882.44 rating

Confidence interval: 865.97 – 898.91

1375 votes

Category
instruction_following
Evaluated configuration
RWKV-4 Raven 14B
Evaluation type
human preference
Arena text preference / mathhuman preference · 2026-10-28 · rating

RWKV-4 Raven 14B

921.91 rating

Confidence interval: 897.82 – 946.01

544 votes

Category
math
Evaluated configuration
RWKV-4 Raven 14B
Evaluation type
human preference
Arena text preference / multi_turnhuman preference · 2026-10-28 · rating

RWKV-4 Raven 14B

893.72 rating

Confidence interval: 866.83 – 920.60

556 votes

Category
multi_turn
Evaluated configuration
RWKV-4 Raven 14B
Evaluation type
human preference
Arena text preference / non_englishhuman preference · 2026-10-28 · rating

RWKV-4 Raven 14B

924.67 rating

Confidence interval: 905.29 – 944.05

848 votes

Category
non_english
Evaluated configuration
RWKV-4 Raven 14B
Evaluation type
human preference
Arena text preference / overallhuman preference · 2026-10-28 · rating

RWKV-4 Raven 14B

948.36 rating

Confidence interval: 937.13 – 959.59

4845 votes

Category
overall
Evaluated configuration
RWKV-4 Raven 14B
Evaluation type
human preference

Run it locally

Inference · memory estimate

RWKV-4 Raven 14B

Not enough verified architecture information for a reliable VRAM estimate.

Capabilities

Model license
Apache 2.0