GeekRadar / The comparison desk

COMPARE.

AI models

alibaba

QwQ 32B

AI model Released Mar 5, 2025

Context window: 131,072 tokens
Maximum output: 8,192 tokens
Open weights: Yes

Capability & preference

General capability · ECI

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

QwQ 32B 137.5^* QwQ-32B · 95% interval 133.1–141.3

Human preference · Arena

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

QwQ 32B 1329.0 QwQ-32B · 95% interval 1324.6–1333.5 · 25,177 votes · 2026-10-28

Model performance

Radar task score: four tests, equal weight. An experimental summary of these tasks, not a universal intelligence rating.

QwQ 32B

Tested configuration Radar task score

Unrated

1/4 core tests · not enough evidence

2 benchmark results for this configuration

Scientific reasoning

GPQA Diamond · 25%

Expert-written science questions; strict multiple-choice scoring. Chance accuracy is 25%.

QwQ 32B 65.3^*% Approx. 95% interval: 59.8–70.9% · 2026-08-28

Novel problem solving

Mystery Game Puzzles · 25%

Best-next-move puzzles in an undisclosed game variant. A narrow reasoning test, not general intelligence.

QwQ 32B Not available

Advanced mathematics

FrontierMath Tiers 1-3 · 25%

Private v2 mathematics problems with Python tools. Earlier versions and Tier 4 are excluded.

QwQ 32B Not available

Factual answers

SimpleQA Verified · 25%

Correct factual answers under a guessing prompt, graded by a model. Not a hallucination or abstention score.

QwQ 32B Not available

Competition mathematics

OTIS Mock AIME 2024-2025 · Separate result

Competition problems with model-extracted answers checked against an exact key. Earlier grading procedures are excluded.

QwQ 32B 59.2^*% Approx. 95% interval: 46.7–71.6% · 2026-08-28

Bars show reported task scores. Whiskers are approximate 95% intervals where standard errors are available. Different reasoning budgets are not equal-cost tests. Missing results are never zero.

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

QwQ 32B

1,378.12 rating

Confidence interval: 1,363.27 – 1,392.97

1532 votes

Oct 28, 2026

Category: chinese
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,332.96 rating

Confidence interval: 1,323.96 – 1,341.96

4017 votes

Oct 28, 2026

Category: coding
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,287.46 rating

Confidence interval: 1,277.68 – 1,297.24

3679 votes

Oct 28, 2026

Category: creative_writing
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,350.64 rating

Confidence interval: 1,344.95 – 1,356.33

12937 votes

Oct 28, 2026

Category: english
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,278.97 rating

Confidence interval: 1,272.52 – 1,285.42

17111 votes

Oct 28, 2026

Category: exclude_ties
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,323.08 rating

Confidence interval: 1,306.68 – 1,339.48

1213 votes

Oct 28, 2026

Category: expert
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,336.83 rating

Confidence interval: 1,304.43 – 1,369.22

316 votes

Oct 28, 2026

Category: french
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,312.22 rating

Confidence interval: 1,289.70 – 1,334.75

670 votes

Oct 28, 2026

Category: german
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,325.36 rating

Confidence interval: 1,318.88 – 1,331.83

8791 votes

Oct 28, 2026

Category: hard_prompts
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,340.81 rating

Confidence interval: 1,332.30 – 1,349.32

4646 votes

Oct 28, 2026

Category: hard_prompts_english
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,319.76 rating

Confidence interval: 1,309.93 – 1,329.60

3475 votes

Oct 28, 2026

Category: industry_business_and_management_and_financial_operations
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,279.22 rating

Confidence interval: 1,270.50 – 1,287.95

4457 votes

Oct 28, 2026

Category: industry_entertainment_and_sports_and_media
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,336.43 rating

Confidence interval: 1,322.01 – 1,350.86

1620 votes

Oct 28, 2026

Category: industry_legal_and_government
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,346.66 rating

Confidence interval: 1,337.78 – 1,355.54

4521 votes

Oct 28, 2026

Category: industry_life_and_physical_and_social_science
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,356.05 rating

Confidence interval: 1,341.78 – 1,370.32

1571 votes

Oct 28, 2026

Category: industry_mathematical
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,339.56 rating

Confidence interval: 1,323.68 – 1,355.45

1350 votes

Oct 28, 2026

Category: industry_medicine_and_healthcare
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,346.88 rating

Confidence interval: 1,339.85 – 1,353.91

7348 votes

Oct 28, 2026

Category: industry_software_and_it_services
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,301.11 rating

Confidence interval: 1,293.51 – 1,308.71

6133 votes

Oct 28, 2026

Category: industry_writing_and_literature_and_language
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,296.84 rating

Confidence interval: 1,289.84 – 1,303.83

7184 votes

Oct 28, 2026

Category: instruction_following
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,262.20 rating

Confidence interval: 1,237.24 – 1,287.16

618 votes

Oct 28, 2026

Category: japanese
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,278.43 rating

Confidence interval: 1,248.22 – 1,308.65

409 votes

Oct 28, 2026

Category: korean
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,307.90 rating

Confidence interval: 1,298.75 – 1,317.05

4052 votes

Oct 28, 2026

Category: longer_query
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,358.91 rating

Confidence interval: 1,345.33 – 1,372.50

1713 votes

Oct 28, 2026

Category: math
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,313.53 rating

Confidence interval: 1,304.01 – 1,323.06

3753 votes

Oct 28, 2026

Category: multi_turn
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,305.27 rating

Confidence interval: 1,299.56 – 1,310.98

12238 votes

Oct 28, 2026

Category: non_english
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,329.04 rating

Confidence interval: 1,324.61 – 1,333.48

25177 votes

Oct 28, 2026

Category: overall
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,287.49 rating

Confidence interval: 1,270.28 – 1,304.70

1248 votes

Oct 28, 2026

Category: polish
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,297.03 rating

Confidence interval: 1,284.60 – 1,309.45

2134 votes

Oct 28, 2026

Category: russian
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,353.56 rating

Confidence interval: 1,324.20 – 1,382.91

396 votes

Oct 28, 2026

Category: spanish
Evaluated configuration: QwQ-32B
Evaluation type: human preference

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

QwQ 32B

1,378.09 rating

Confidence interval: 1,363.24 – 1,392.93

1533 votes

Oct 15, 2026

Category: chinese
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,333.02 rating

Confidence interval: 1,324.02 – 1,342.03

4016 votes

Oct 15, 2026

Category: coding
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,287.54 rating

Confidence interval: 1,277.76 – 1,297.31

3680 votes

Oct 15, 2026

Category: creative_writing
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,350.59 rating

Confidence interval: 1,344.90 – 1,356.28

12938 votes

Oct 15, 2026

Category: english
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,279.07 rating

Confidence interval: 1,272.62 – 1,285.52

17112 votes

Oct 15, 2026

Category: exclude_ties
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,323.11 rating

Confidence interval: 1,306.71 – 1,339.51

1213 votes

Oct 15, 2026

Category: expert
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,336.40 rating

Confidence interval: 1,303.99 – 1,368.80

316 votes

Oct 15, 2026

Category: french
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,312.77 rating

Confidence interval: 1,290.27 – 1,335.27

671 votes

Oct 15, 2026

Category: german
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,325.41 rating

Confidence interval: 1,318.94 – 1,331.88

8791 votes

Oct 15, 2026

Category: hard_prompts
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,340.65 rating

Confidence interval: 1,332.14 – 1,349.17

4647 votes

Oct 15, 2026

Category: hard_prompts_english
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,319.83 rating

Confidence interval: 1,309.99 – 1,329.66

3475 votes

Oct 15, 2026

Category: industry_business_and_management_and_financial_operations
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,279.36 rating

Confidence interval: 1,270.64 – 1,288.08

4459 votes

Oct 15, 2026

Category: industry_entertainment_and_sports_and_media
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,336.18 rating

Confidence interval: 1,321.76 – 1,350.60

1620 votes

Oct 15, 2026

Category: industry_legal_and_government
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,346.73 rating

Confidence interval: 1,337.85 – 1,355.61

4521 votes

Oct 15, 2026

Category: industry_life_and_physical_and_social_science
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,356.08 rating

Confidence interval: 1,341.81 – 1,370.34

1571 votes

Oct 15, 2026

Category: industry_mathematical
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,339.87 rating

Confidence interval: 1,323.98 – 1,355.76

1349 votes

Oct 15, 2026

Category: industry_medicine_and_healthcare
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,346.93 rating

Confidence interval: 1,339.90 – 1,353.95

7348 votes

Oct 15, 2026

Category: industry_software_and_it_services
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,301.09 rating

Confidence interval: 1,293.49 – 1,308.69

6135 votes

Oct 15, 2026

Category: industry_writing_and_literature_and_language
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,296.86 rating

Confidence interval: 1,289.87 – 1,303.85

7184 votes

Oct 15, 2026

Category: instruction_following
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,262.04 rating

Confidence interval: 1,237.10 – 1,286.98

618 votes

Oct 15, 2026

Category: japanese
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,278.57 rating

Confidence interval: 1,248.37 – 1,308.78

409 votes

Oct 15, 2026

Category: korean
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,307.94 rating

Confidence interval: 1,298.79 – 1,317.09

4052 votes

Oct 15, 2026

Category: longer_query
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,358.92 rating

Confidence interval: 1,345.34 – 1,372.51

1713 votes

Oct 15, 2026

Category: math
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,313.63 rating

Confidence interval: 1,304.10 – 1,323.15

3754 votes

Oct 15, 2026

Category: multi_turn
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,305.44 rating

Confidence interval: 1,299.73 – 1,311.15

12238 votes

Oct 15, 2026

Category: non_english
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,329.08 rating

Confidence interval: 1,324.65 – 1,333.51

25178 votes

Oct 15, 2026

Category: overall
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,287.86 rating

Confidence interval: 1,270.66 – 1,305.07

1247 votes

Oct 15, 2026

Category: polish
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,297.14 rating

Confidence interval: 1,284.72 – 1,309.56

2134 votes

Oct 15, 2026

Category: russian
Evaluated configuration: qwq-32b
Evaluation type: human preference

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

QwQ 32B

1,353.44 rating

Confidence interval: 1,324.09 – 1,382.80

396 votes

Oct 15, 2026

Category: spanish
Evaluated configuration: qwq-32b
Evaluation type: human preference

Run it locally

Inference · memory estimate

QwQ 32B

Not enough verified architecture information for a reliable VRAM estimate.

API cost

USD / month

Input tokens / monthOutput tokens / month

Standard text-token estimate. Excludes taxes, caching, batch discounts, reasoning and multimodal surcharges. Provider limits and regional availability may differ.

QwQ 32B

Provider offer

Capabilities

Context window: 131,072 tokens
Maximum output: 8,192 tokens
Reasoning: Yes
Tool calling: Yes
Open weights: Yes
Knowledge cutoff: 2024-04
Model license: Apache 2.0

Input modalities

text

Output modalities

text

* Provisional data