GeekRadar / The comparison desk
COMPARE.
AI models
Not reported

Zephyr 7B Alpha

Model evaluation

Capability & preference

Human preference · Arena

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

Zephyr 7B Alpha 1058.3 Zephyr 7B Alpha · 95% interval 1042.7–1074.0 · 1,785 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 / codinghuman preference · 2026-10-28 · rating

Zephyr 7B Alpha

1,052.46 rating

Confidence interval: 1,013.49 – 1,091.44

201 votes

Category
coding
Evaluated configuration
Zephyr 7B Alpha
Evaluation type
human preference
Arena text preference / creative_writinghuman preference · 2026-10-28 · rating

Zephyr 7B Alpha

1,068.44 rating

Confidence interval: 1,038.52 – 1,098.36

399 votes

Category
creative_writing
Evaluated configuration
Zephyr 7B Alpha
Evaluation type
human preference
Arena text preference / englishhuman preference · 2026-10-28 · rating

Zephyr 7B Alpha

1,091.64 rating

Confidence interval: 1,074.28 – 1,108.99

1415 votes

Category
english
Evaluated configuration
Zephyr 7B Alpha
Evaluation type
human preference
Arena text preference / exclude_tieshuman preference · 2026-10-28 · rating

Zephyr 7B Alpha

875.22 rating

Confidence interval: 852.82 – 897.62

1218 votes

Category
exclude_ties
Evaluated configuration
Zephyr 7B Alpha
Evaluation type
human preference
Arena text preference / hard_promptshuman preference · 2026-10-28 · rating

Zephyr 7B Alpha

1,023.71 rating

Confidence interval: 995.00 – 1,052.42

333 votes

Category
hard_prompts
Evaluated configuration
Zephyr 7B Alpha
Evaluation type
human preference
Arena text preference / hard_prompts_englishhuman preference · 2026-10-28 · rating

Zephyr 7B Alpha

1,041.93 rating

Confidence interval: 1,009.55 – 1,074.30

278 votes

Category
hard_prompts_english
Evaluated configuration
Zephyr 7B Alpha
Evaluation type
human preference
Arena text preference / industry_entertainment_and_sports_and_mediahuman preference · 2026-10-28 · rating

Zephyr 7B Alpha

1,048.18 rating

Confidence interval: 1,016.46 – 1,079.90

329 votes

Category
industry_entertainment_and_sports_and_media
Evaluated configuration
Zephyr 7B Alpha
Evaluation type
human preference
Arena text preference / industry_life_and_physical_and_social_sciencehuman preference · 2026-10-28 · rating

Zephyr 7B Alpha

1,027.97 rating

Confidence interval: 994.00 – 1,061.94

294 votes

Category
industry_life_and_physical_and_social_science
Evaluated configuration
Zephyr 7B Alpha
Evaluation type
human preference
Arena text preference / industry_software_and_it_serviceshuman preference · 2026-10-28 · rating

Zephyr 7B Alpha

1,053.17 rating

Confidence interval: 1,024.85 – 1,081.49

406 votes

Category
industry_software_and_it_services
Evaluated configuration
Zephyr 7B Alpha
Evaluation type
human preference
Arena text preference / industry_writing_and_literature_and_languagehuman preference · 2026-10-28 · rating

Zephyr 7B Alpha

1,081.55 rating

Confidence interval: 1,056.32 – 1,106.77

569 votes

Category
industry_writing_and_literature_and_language
Evaluated configuration
Zephyr 7B Alpha
Evaluation type
human preference
Arena text preference / instruction_followinghuman preference · 2026-10-28 · rating

Zephyr 7B Alpha

1,034.22 rating

Confidence interval: 1,009.96 – 1,058.48

534 votes

Category
instruction_following
Evaluated configuration
Zephyr 7B Alpha
Evaluation type
human preference
Arena text preference / multi_turnhuman preference · 2026-10-28 · rating

Zephyr 7B Alpha

1,017.59 rating

Confidence interval: 983.15 – 1,052.03

283 votes

Category
multi_turn
Evaluated configuration
Zephyr 7B Alpha
Evaluation type
human preference
Arena text preference / non_englishhuman preference · 2026-10-28 · rating

Zephyr 7B Alpha

987.73 rating

Confidence interval: 959.33 – 1,016.13

370 votes

Category
non_english
Evaluated configuration
Zephyr 7B Alpha
Evaluation type
human preference
Arena text preference / overallhuman preference · 2026-10-28 · rating

Zephyr 7B Alpha

1,058.34 rating

Confidence interval: 1,042.69 – 1,074.00

1785 votes

Category
overall
Evaluated configuration
Zephyr 7B Alpha
Evaluation type
human preference

Run it locally

Inference · memory estimate

Zephyr 7B Alpha

Not enough verified architecture information for a reliable VRAM estimate.

Capabilities

Model license
MIT