geekradar
Int
Explore GeekRadar

More to explore

Good gear All topics About us Say hello
GeekRadar / The comparison desk
COMPARE.
AI models
alibaba

Qwen2.5-72B-Instruct

AI model

Context window
32,768 tokens
Total parameters
72.71 B
Open weights
Yes

Capability & preference

Human preference · Arena

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

Qwen2.5-72B-Instruct 1269.0 Qwen2.5-72B-Instruct · 95% interval 1264.9–1273.1 · 39,406 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

Qwen2.5-72B-Instruct

1,272.33 rating

Confidence interval: 1,261.32 – 1,283.34

3329 votes

Category
chinese
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / codinghuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,292.16 rating

Confidence interval: 1,284.56 – 1,299.76

6688 votes

Category
coding
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / creative_writinghuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,220.96 rating

Confidence interval: 1,212.54 – 1,229.38

5684 votes

Category
creative_writing
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / englishhuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,282.61 rating

Confidence interval: 1,277.46 – 1,287.75

20561 votes

Category
english
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / exclude_tieshuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,189.01 rating

Confidence interval: 1,182.63 – 1,195.40

24444 votes

Category
exclude_ties
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / experthuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,245.06 rating

Confidence interval: 1,232.94 – 1,257.19

2397 votes

Category
expert
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / frenchhuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,280.54 rating

Confidence interval: 1,247.84 – 1,313.25

340 votes

Category
french
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / germanhuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,233.50 rating

Confidence interval: 1,214.23 – 1,252.76

926 votes

Category
german
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / hard_promptshuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,270.48 rating

Confidence interval: 1,264.14 – 1,276.82

10545 votes

Category
hard_prompts
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / hard_prompts_englishhuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,279.36 rating

Confidence interval: 1,271.60 – 1,287.12

6208 votes

Category
hard_prompts_english
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / industry_business_and_management_and_financial_operationshuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,252.81 rating

Confidence interval: 1,243.54 – 1,262.07

4656 votes

Category
industry_business_and_management_and_financial_operations
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / industry_entertainment_and_sports_and_mediahuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,210.93 rating

Confidence interval: 1,202.92 – 1,218.95

6207 votes

Category
industry_entertainment_and_sports_and_media
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / industry_legal_and_governmenthuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,287.96 rating

Confidence interval: 1,275.16 – 1,300.76

2386 votes

Category
industry_legal_and_government
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / industry_life_and_physical_and_social_sciencehuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,274.85 rating

Confidence interval: 1,266.98 – 1,282.72

6932 votes

Category
industry_life_and_physical_and_social_science
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / industry_mathematicalhuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,290.44 rating

Confidence interval: 1,281.62 – 1,299.26

4565 votes

Category
industry_mathematical
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / industry_medicine_and_healthcarehuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,258.34 rating

Confidence interval: 1,244.36 – 1,272.32

1956 votes

Category
industry_medicine_and_healthcare
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / industry_software_and_it_serviceshuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,292.06 rating

Confidence interval: 1,285.48 – 1,298.63

10448 votes

Category
industry_software_and_it_services
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / industry_writing_and_literature_and_languagehuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,246.84 rating

Confidence interval: 1,240.10 – 1,253.59

10827 votes

Category
industry_writing_and_literature_and_language
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / instruction_followinghuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,253.91 rating

Confidence interval: 1,248.27 – 1,259.55

16363 votes

Category
instruction_following
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / japanesehuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,179.91 rating

Confidence interval: 1,158.54 – 1,201.27

788 votes

Category
japanese
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / koreanhuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,187.99 rating

Confidence interval: 1,160.21 – 1,215.78

527 votes

Category
korean
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / longer_queryhuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,281.36 rating

Confidence interval: 1,273.30 – 1,289.42

6008 votes

Category
longer_query
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / mathhuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,282.57 rating

Confidence interval: 1,274.32 – 1,290.82

5415 votes

Category
math
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / multi_turnhuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,271.77 rating

Confidence interval: 1,263.89 – 1,279.65

7250 votes

Category
multi_turn
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / non_englishhuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,251.72 rating

Confidence interval: 1,246.28 – 1,257.16

18845 votes

Category
non_english
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / overallhuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,269.00 rating

Confidence interval: 1,264.93 – 1,273.06

39406 votes

Category
overall
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / russianhuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,263.73 rating

Confidence interval: 1,255.27 – 1,272.20

5776 votes

Category
russian
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / spanishhuman preference · 2026-10-28 · rating

Qwen2.5-72B-Instruct

1,255.47 rating

Confidence interval: 1,224.53 – 1,286.40

347 votes

Category
spanish
Evaluated configuration
Qwen2.5-72B-Instruct
Evaluation type
human preference
Arena text preference / chinesehuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,272.38 rating

Confidence interval: 1,261.37 – 1,283.39

3329 votes

Category
chinese
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / codinghuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,292.26 rating

Confidence interval: 1,284.66 – 1,299.86

6688 votes

Category
coding
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / creative_writinghuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,221.12 rating

Confidence interval: 1,212.69 – 1,229.54

5684 votes

Category
creative_writing
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / englishhuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,282.55 rating

Confidence interval: 1,277.41 – 1,287.69

20561 votes

Category
english
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / exclude_tieshuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,189.02 rating

Confidence interval: 1,182.64 – 1,195.41

24444 votes

Category
exclude_ties
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / experthuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,245.10 rating

Confidence interval: 1,232.97 – 1,257.24

2397 votes

Category
expert
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / frenchhuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,280.23 rating

Confidence interval: 1,247.52 – 1,312.95

340 votes

Category
french
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / germanhuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,233.49 rating

Confidence interval: 1,214.27 – 1,252.71

926 votes

Category
german
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / hard_promptshuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,270.47 rating

Confidence interval: 1,264.13 – 1,276.81

10545 votes

Category
hard_prompts
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / hard_prompts_englishhuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,279.25 rating

Confidence interval: 1,271.49 – 1,287.01

6208 votes

Category
hard_prompts_english
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / industry_business_and_management_and_financial_operationshuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,252.84 rating

Confidence interval: 1,243.58 – 1,262.11

4656 votes

Category
industry_business_and_management_and_financial_operations
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / industry_entertainment_and_sports_and_mediahuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,210.99 rating

Confidence interval: 1,202.98 – 1,219.01

6207 votes

Category
industry_entertainment_and_sports_and_media
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / industry_legal_and_governmenthuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,287.71 rating

Confidence interval: 1,274.91 – 1,300.51

2386 votes

Category
industry_legal_and_government
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / industry_life_and_physical_and_social_sciencehuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,274.83 rating

Confidence interval: 1,266.96 – 1,282.70

6932 votes

Category
industry_life_and_physical_and_social_science
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / industry_mathematicalhuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,290.43 rating

Confidence interval: 1,281.62 – 1,299.25

4565 votes

Category
industry_mathematical
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / industry_medicine_and_healthcarehuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,258.40 rating

Confidence interval: 1,244.42 – 1,272.38

1956 votes

Category
industry_medicine_and_healthcare
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / industry_software_and_it_serviceshuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,292.14 rating

Confidence interval: 1,285.57 – 1,298.72

10448 votes

Category
industry_software_and_it_services
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / industry_writing_and_literature_and_languagehuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,246.91 rating

Confidence interval: 1,240.16 – 1,253.65

10827 votes

Category
industry_writing_and_literature_and_language
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / instruction_followinghuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,253.92 rating

Confidence interval: 1,248.28 – 1,259.56

16363 votes

Category
instruction_following
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / japanesehuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,179.74 rating

Confidence interval: 1,158.43 – 1,201.04

788 votes

Category
japanese
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / koreanhuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,188.04 rating

Confidence interval: 1,160.27 – 1,215.81

527 votes

Category
korean
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / longer_queryhuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,281.40 rating

Confidence interval: 1,273.34 – 1,289.46

6008 votes

Category
longer_query
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / mathhuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,282.53 rating

Confidence interval: 1,274.28 – 1,290.78

5415 votes

Category
math
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / multi_turnhuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,271.77 rating

Confidence interval: 1,263.89 – 1,279.65

7250 votes

Category
multi_turn
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / non_englishhuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,251.81 rating

Confidence interval: 1,246.37 – 1,257.25

18845 votes

Category
non_english
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / overallhuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,268.98 rating

Confidence interval: 1,264.92 – 1,273.04

39406 votes

Category
overall
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / russianhuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,263.89 rating

Confidence interval: 1,255.42 – 1,272.36

5776 votes

Category
russian
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference
Arena text preference / spanishhuman preference · 2026-10-15 · rating

Qwen2.5-72B-Instruct

1,255.39 rating

Confidence interval: 1,224.44 – 1,286.34

347 votes

Category
spanish
Evaluated configuration
qwen2.5-72b-instruct
Evaluation type
human preference

Run it locally

Inference · memory estimate

Qwen2.5-72B-Instruct

72.71 B total parameters · qwen2 · BF16

Published checkpoint tensor payload: 135.4 GiB. This includes stored copies and buffers; it is separate from the runtime estimates below.

Estimated accelerator memory42.2 GiB
Weights
37.2 GiB
KV cache
1.3 GiB
Runtime allowance
3.7 GiB

One user, full model in accelerator memory.

GPUs with memory headroom

Local AI systems with memory headroom

Shared-memory systems use a planning budget of 75% of installed memory, leaving 25% for the OS and other allocations. This is an assumption, not a measured GPU allocation. Firmware and driver limits may be lower and may require configuration. Memory matches do not verify a runtime, quantized model artifact or inference speed.

Includes FP16 KV cache, prompt plus generated tokens, quantization metadata and a runtime allowance. GPU matches reserve 10% of VRAM. Memory fit is not a speed or software-compatibility guarantee; quantized weights and runtime support must be available. Cache uses a conservative full-attention upper bound.

Capabilities

Context window
32,768 tokens
Open weights
Yes
Model license
other
Total parameters
72.71 B
Checkpoint tensor payload
135.43 GiB

Input modalities

  • text

Output modalities

  • text

Processing & platform

Architecture
qwen2