GeekRadar / The comparison desk

COMPARE.

AI models

microsoft

Phi-3-mini-4k-instruct

AI model

Context window: 4,096 tokens
Open weights: Yes

Capability & preference

Human preference · Arena

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

Phi-3-mini-4k-instruct 1073.3 Phi-3 Mini 4K Instruct · 95% interval 1067.0–1079.6 · 20,118 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.

Test version

Arena text preference / chinesehuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,020.95 rating

Confidence interval: 1,006.53 – 1,035.38

2296 votes

Oct 28, 2026

Category: chinese
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / codinghuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,092.33 rating

Confidence interval: 1,080.74 – 1,103.92

3449 votes

Oct 28, 2026

Category: coding
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / creative_writinghuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,036.63 rating

Confidence interval: 1,024.02 – 1,049.23

2812 votes

Oct 28, 2026

Category: creative_writing
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / englishhuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,117.22 rating

Confidence interval: 1,109.13 – 1,125.31

10322 votes

Oct 28, 2026

Category: english
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / exclude_tieshuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

871.96 rating

Confidence interval: 860.73 – 883.20

12861 votes

Oct 28, 2026

Category: exclude_ties
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / experthuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,045.13 rating

Confidence interval: 1,025.82 – 1,064.44

936 votes

Oct 28, 2026

Category: expert
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / frenchhuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,076.40 rating

Confidence interval: 1,040.04 – 1,112.76

307 votes

Oct 28, 2026

Category: french
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / germanhuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,043.89 rating

Confidence interval: 1,019.92 – 1,067.87

699 votes

Oct 28, 2026

Category: german
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / hard_promptshuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,072.06 rating

Confidence interval: 1,062.61 – 1,081.50

5872 votes

Oct 28, 2026

Category: hard_prompts
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / hard_prompts_englishhuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,097.70 rating

Confidence interval: 1,086.22 – 1,109.19

3447 votes

Oct 28, 2026

Category: hard_prompts_english
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / industry_business_and_management_and_financial_operationshuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,032.04 rating

Confidence interval: 1,018.57 – 1,045.52

2483 votes

Oct 28, 2026

Category: industry_business_and_management_and_financial_operations
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / industry_entertainment_and_sports_and_mediahuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,018.84 rating

Confidence interval: 1,006.72 – 1,030.97

2995 votes

Oct 28, 2026

Category: industry_entertainment_and_sports_and_media
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / industry_legal_and_governmenthuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,071.05 rating

Confidence interval: 1,051.92 – 1,090.17

1126 votes

Oct 28, 2026

Category: industry_legal_and_government
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / industry_life_and_physical_and_social_sciencehuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,061.86 rating

Confidence interval: 1,049.38 – 1,074.35

3517 votes

Oct 28, 2026

Category: industry_life_and_physical_and_social_science
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / industry_mathematicalhuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,109.17 rating

Confidence interval: 1,096.15 – 1,122.20

2198 votes

Oct 28, 2026

Category: industry_mathematical
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / industry_medicine_and_healthcarehuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,029.92 rating

Confidence interval: 1,009.59 – 1,050.24

1078 votes

Oct 28, 2026

Category: industry_medicine_and_healthcare
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / industry_software_and_it_serviceshuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,081.45 rating

Confidence interval: 1,071.65 – 1,091.24

5715 votes

Oct 28, 2026

Category: industry_software_and_it_services
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / industry_writing_and_literature_and_languagehuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,054.32 rating

Confidence interval: 1,044.03 – 1,064.60

5207 votes

Oct 28, 2026

Category: industry_writing_and_literature_and_language
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / instruction_followinghuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,053.12 rating

Confidence interval: 1,044.41 – 1,061.83

7636 votes

Oct 28, 2026

Category: instruction_following
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / japanesehuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

935.13 rating

Confidence interval: 900.82 – 969.44

384 votes

Oct 28, 2026

Category: japanese
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / koreanhuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

905.41 rating

Confidence interval: 880.50 – 930.32

860 votes

Oct 28, 2026

Category: korean
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / longer_queryhuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,043.70 rating

Confidence interval: 1,028.78 – 1,058.62

1706 votes

Oct 28, 2026

Category: longer_query
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / mathhuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,111.40 rating

Confidence interval: 1,099.18 – 1,123.61

2564 votes

Oct 28, 2026

Category: math
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / multi_turnhuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,018.25 rating

Confidence interval: 1,004.90 – 1,031.59

2939 votes

Oct 28, 2026

Category: multi_turn
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / non_englishhuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,020.54 rating

Confidence interval: 1,012.31 – 1,028.77

9796 votes

Oct 28, 2026

Category: non_english
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / overallhuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,073.28 rating

Confidence interval: 1,066.99 – 1,079.58

20118 votes

Oct 28, 2026

Category: overall
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / russianhuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,022.02 rating

Confidence interval: 1,008.96 – 1,035.08

2751 votes

Oct 28, 2026

Category: russian
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

Arena text preference / spanishhuman preference · 2026-10-28 · rating

Phi-3-mini-4k-instruct

1,084.79 rating

Confidence interval: 1,052.25 – 1,117.33

410 votes

Oct 28, 2026

Category: spanish
Evaluated configuration: Phi-3 Mini 4K Instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,021.00 rating

Confidence interval: 1,006.57 – 1,035.43

2296 votes

Oct 15, 2026

Category: chinese
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,092.43 rating

Confidence interval: 1,080.83 – 1,104.02

3449 votes

Oct 15, 2026

Category: coding
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,036.78 rating

Confidence interval: 1,024.17 – 1,049.39

2812 votes

Oct 15, 2026

Category: creative_writing
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,117.17 rating

Confidence interval: 1,109.08 – 1,125.26

10322 votes

Oct 15, 2026

Category: english
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

871.97 rating

Confidence interval: 860.74 – 883.21

12861 votes

Oct 15, 2026

Category: exclude_ties
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,045.17 rating

Confidence interval: 1,025.86 – 1,064.49

936 votes

Oct 15, 2026

Category: expert
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,076.08 rating

Confidence interval: 1,039.70 – 1,112.47

307 votes

Oct 15, 2026

Category: french
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,043.89 rating

Confidence interval: 1,019.96 – 1,067.82

699 votes

Oct 15, 2026

Category: german
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,072.05 rating

Confidence interval: 1,062.60 – 1,081.49

5872 votes

Oct 15, 2026

Category: hard_prompts
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,097.59 rating

Confidence interval: 1,086.10 – 1,109.08

3447 votes

Oct 15, 2026

Category: hard_prompts_english
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,032.08 rating

Confidence interval: 1,018.61 – 1,045.56

2483 votes

Oct 15, 2026

Category: industry_business_and_management_and_financial_operations
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,018.90 rating

Confidence interval: 1,006.77 – 1,031.03

2995 votes

Oct 15, 2026

Category: industry_entertainment_and_sports_and_media
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,070.79 rating

Confidence interval: 1,051.66 – 1,089.92

1126 votes

Oct 15, 2026

Category: industry_legal_and_government
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,061.84 rating

Confidence interval: 1,049.35 – 1,074.33

3517 votes

Oct 15, 2026

Category: industry_life_and_physical_and_social_science
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,109.17 rating

Confidence interval: 1,096.14 – 1,122.19

2198 votes

Oct 15, 2026

Category: industry_mathematical
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,029.97 rating

Confidence interval: 1,009.65 – 1,050.30

1078 votes

Oct 15, 2026

Category: industry_medicine_and_healthcare
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,081.53 rating

Confidence interval: 1,071.73 – 1,091.33

5715 votes

Oct 15, 2026

Category: industry_software_and_it_services
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,054.38 rating

Confidence interval: 1,044.09 – 1,064.67

5207 votes

Oct 15, 2026

Category: industry_writing_and_literature_and_language
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,053.13 rating

Confidence interval: 1,044.42 – 1,061.85

7636 votes

Oct 15, 2026

Category: instruction_following
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

934.96 rating

Confidence interval: 900.69 – 969.23

384 votes

Oct 15, 2026

Category: japanese
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

905.45 rating

Confidence interval: 880.56 – 930.34

860 votes

Oct 15, 2026

Category: korean
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,043.74 rating

Confidence interval: 1,028.82 – 1,058.67

1706 votes

Oct 15, 2026

Category: longer_query
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,111.36 rating

Confidence interval: 1,099.14 – 1,123.57

2564 votes

Oct 15, 2026

Category: math
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,018.24 rating

Confidence interval: 1,004.90 – 1,031.59

2939 votes

Oct 15, 2026

Category: multi_turn
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,020.64 rating

Confidence interval: 1,012.41 – 1,028.87

9796 votes

Oct 15, 2026

Category: non_english
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,073.27 rating

Confidence interval: 1,066.97 – 1,079.56

20118 votes

Oct 15, 2026

Category: overall
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,022.17 rating

Confidence interval: 1,009.10 – 1,035.23

2751 votes

Oct 15, 2026

Category: russian
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

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

Phi-3-mini-4k-instruct

1,084.71 rating

Confidence interval: 1,052.14 – 1,117.28

410 votes

Oct 15, 2026

Category: spanish
Evaluated configuration: phi-3-mini-4k-instruct
Evaluation type: human preference

Run it locally

Inference · memory estimate

Phi-3-mini-4k-instruct

7.1 GiB checkpoint tensor payload · 3.82 B stored tensor elements · BF16

Stored tensors can include buffers and repeated shared weights. A verified unique parameter count is unavailable. Checkpoint size is not the memory required to run the model.

Capabilities

Context window: 4,096 tokens
Open weights: Yes
Model license: mit
Checkpoint tensor payload: 7.12 GiB

Input modalities

text

Output modalities

text

Processing & platform

Architecture: phi3