GeekRadar / The comparison desk
COMPARE.
AI models
meta

Llama 3.1 405B Instruct (FP8)

Model evaluation

Capability & preference

Human preference · Arena

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

Llama 3.1 405B Instruct (FP8) 1281.8 Llama 3.1 405B Instruct (FP8) · 95% interval 1278.2–1285.3 · 59,656 votes · 2026-11-06

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 · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,234.50 rating

Confidence interval: 1,224.78 – 1,244.22

5226 votes

Category
chinese
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / codinghuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,283.26 rating

Confidence interval: 1,276.17 – 1,290.34

9714 votes

Category
coding
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / creative_writinghuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,261.64 rating

Confidence interval: 1,253.96 – 1,269.32

8948 votes

Category
creative_writing
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / englishhuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,307.92 rating

Confidence interval: 1,303.16 – 1,312.67

32442 votes

Category
english
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / exclude_tieshuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,208.74 rating

Confidence interval: 1,203.02 – 1,214.46

38128 votes

Category
exclude_ties
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / experthuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,242.82 rating

Confidence interval: 1,231.45 – 1,254.20

3123 votes

Category
expert
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / frenchhuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,271.05 rating

Confidence interval: 1,243.42 – 1,298.69

545 votes

Category
french
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / germanhuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,250.48 rating

Confidence interval: 1,234.15 – 1,266.81

1587 votes

Category
german
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / hard_promptshuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,263.22 rating

Confidence interval: 1,257.43 – 1,269.01

16228 votes

Category
hard_prompts
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / hard_prompts_englishhuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,283.07 rating

Confidence interval: 1,276.04 – 1,290.11

9964 votes

Category
hard_prompts_english
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / industry_business_and_management_and_financial_operationshuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,247.33 rating

Confidence interval: 1,238.87 – 1,255.79

6667 votes

Category
industry_business_and_management_and_financial_operations
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / industry_entertainment_and_sports_and_mediahuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,249.97 rating

Confidence interval: 1,242.70 – 1,257.25

9523 votes

Category
industry_entertainment_and_sports_and_media
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / industry_legal_and_governmenthuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,303.50 rating

Confidence interval: 1,291.94 – 1,315.07

3383 votes

Category
industry_legal_and_government
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / industry_life_and_physical_and_social_sciencehuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,284.23 rating

Confidence interval: 1,276.94 – 1,291.52

10001 votes

Category
industry_life_and_physical_and_social_science
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / industry_mathematicalhuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,276.12 rating

Confidence interval: 1,268.12 – 1,284.12

7242 votes

Category
industry_mathematical
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / industry_medicine_and_healthcarehuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,270.58 rating

Confidence interval: 1,258.32 – 1,282.84

3064 votes

Category
industry_medicine_and_healthcare
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / industry_software_and_it_serviceshuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,280.71 rating

Confidence interval: 1,274.68 – 1,286.73

15817 votes

Category
industry_software_and_it_services
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / industry_writing_and_literature_and_languagehuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,274.10 rating

Confidence interval: 1,267.94 – 1,280.26

15999 votes

Category
industry_writing_and_literature_and_language
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / instruction_followinghuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,258.45 rating

Confidence interval: 1,253.19 – 1,263.72

23585 votes

Category
instruction_following
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / japanesehuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,171.39 rating

Confidence interval: 1,154.53 – 1,188.24

1606 votes

Category
japanese
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / koreanhuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,184.18 rating

Confidence interval: 1,161.96 – 1,206.39

924 votes

Category
korean
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / longer_queryhuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,259.24 rating

Confidence interval: 1,251.71 – 1,266.77

8047 votes

Category
longer_query
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / mathhuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,281.32 rating

Confidence interval: 1,273.76 – 1,288.88

8482 votes

Category
math
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / multi_turnhuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,287.19 rating

Confidence interval: 1,279.99 – 1,294.39

11136 votes

Category
multi_turn
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / non_englishhuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,247.66 rating

Confidence interval: 1,242.72 – 1,252.61

27214 votes

Category
non_english
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / overallhuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,281.76 rating

Confidence interval: 1,278.19 – 1,285.33

59656 votes

Category
overall
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / russianhuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,265.19 rating

Confidence interval: 1,257.27 – 1,273.10

7673 votes

Category
russian
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference
Arena text preference / spanishhuman preference · 2026-11-06 · rating

Llama 3.1 405B Instruct (FP8)

1,253.06 rating

Confidence interval: 1,227.82 – 1,278.29

652 votes

Category
spanish
Evaluated configuration
Llama 3.1 405B Instruct (FP8)
Evaluation type
human preference

Run it locally

Inference · memory estimate

Llama 3.1 405B Instruct (FP8)

Not enough verified architecture information for a reliable VRAM estimate.

Capabilities

Model license
Llama 3.1 Community