geekradar
Int
Explore GeekRadar

More to explore

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

deepseek-llm-67b-chat

AI model Released

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.

deepseek-llm-67b-chat 1105.0 DeepSeek LLM 67B Chat · 95% interval 1093.5–1116.5 · 4,932 votes · 2026-10-28

Model performance

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

deepseek-llm-67b-chat

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%.

deepseek-llm-67b-chat 24.6*% Approx. 95% interval: 22.0–27.3% · Older result

Novel problem solving

Mystery Game Puzzles · 25%

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

deepseek-llm-67b-chat Not available

Advanced mathematics

FrontierMath Tiers 1-3 · 25%

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

deepseek-llm-67b-chat Not available

Factual answers

SimpleQA Verified · 25%

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

deepseek-llm-67b-chat 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.

deepseek-llm-67b-chat 0.8*% Approx. 95% interval: 0.0–1.8%

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.

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

deepseek-llm-67b-chat

1,131.75 rating

Confidence interval: 1,092.42 – 1,171.07

211 votes

Category
chinese
Evaluated configuration
DeepSeek LLM 67B Chat
Evaluation type
human preference
Arena text preference / codinghuman preference · 2026-10-28 · rating

deepseek-llm-67b-chat

1,095.28 rating

Confidence interval: 1,071.79 – 1,118.78

649 votes

Category
coding
Evaluated configuration
DeepSeek LLM 67B Chat
Evaluation type
human preference
Arena text preference / creative_writinghuman preference · 2026-10-28 · rating

deepseek-llm-67b-chat

1,067.12 rating

Confidence interval: 1,045.27 – 1,088.96

789 votes

Category
creative_writing
Evaluated configuration
DeepSeek LLM 67B Chat
Evaluation type
human preference
Arena text preference / englishhuman preference · 2026-10-28 · rating

deepseek-llm-67b-chat

1,127.06 rating

Confidence interval: 1,114.30 – 1,139.83

3820 votes

Category
english
Evaluated configuration
DeepSeek LLM 67B Chat
Evaluation type
human preference
Arena text preference / exclude_tieshuman preference · 2026-10-28 · rating

deepseek-llm-67b-chat

942.35 rating

Confidence interval: 925.68 – 959.02

3443 votes

Category
exclude_ties
Evaluated configuration
DeepSeek LLM 67B Chat
Evaluation type
human preference
Arena text preference / hard_promptshuman preference · 2026-10-28 · rating

deepseek-llm-67b-chat

1,069.68 rating

Confidence interval: 1,051.34 – 1,088.03

1120 votes

Category
hard_prompts
Evaluated configuration
DeepSeek LLM 67B Chat
Evaluation type
human preference
Arena text preference / hard_prompts_englishhuman preference · 2026-10-28 · rating

deepseek-llm-67b-chat

1,085.30 rating

Confidence interval: 1,065.45 – 1,105.16

946 votes

Category
hard_prompts_english
Evaluated configuration
DeepSeek LLM 67B Chat
Evaluation type
human preference
Arena text preference / industry_business_and_management_and_financial_operationshuman preference · 2026-10-28 · rating

deepseek-llm-67b-chat

1,047.58 rating

Confidence interval: 1,019.33 – 1,075.83

441 votes

Category
industry_business_and_management_and_financial_operations
Evaluated configuration
DeepSeek LLM 67B Chat
Evaluation type
human preference
Arena text preference / industry_entertainment_and_sports_and_mediahuman preference · 2026-10-28 · rating

deepseek-llm-67b-chat

1,061.61 rating

Confidence interval: 1,040.13 – 1,083.09

802 votes

Category
industry_entertainment_and_sports_and_media
Evaluated configuration
DeepSeek LLM 67B Chat
Evaluation type
human preference
Arena text preference / industry_legal_and_governmenthuman preference · 2026-10-28 · rating

deepseek-llm-67b-chat

1,089.30 rating

Confidence interval: 1,054.49 – 1,124.11

300 votes

Category
industry_legal_and_government
Evaluated configuration
DeepSeek LLM 67B Chat
Evaluation type
human preference
Arena text preference / industry_life_and_physical_and_social_sciencehuman preference · 2026-10-28 · rating

deepseek-llm-67b-chat

1,065.79 rating

Confidence interval: 1,044.01 – 1,087.56

876 votes

Category
industry_life_and_physical_and_social_science
Evaluated configuration
DeepSeek LLM 67B Chat
Evaluation type
human preference
Arena text preference / industry_mathematicalhuman preference · 2026-10-28 · rating

deepseek-llm-67b-chat

1,093.04 rating

Confidence interval: 1,066.52 – 1,119.56

484 votes

Category
industry_mathematical
Evaluated configuration
DeepSeek LLM 67B Chat
Evaluation type
human preference
Arena text preference / industry_medicine_and_healthcarehuman preference · 2026-10-28 · rating

deepseek-llm-67b-chat

1,061.55 rating

Confidence interval: 1,032.95 – 1,090.16

385 votes

Category
industry_medicine_and_healthcare
Evaluated configuration
DeepSeek LLM 67B Chat
Evaluation type
human preference
Arena text preference / industry_software_and_it_serviceshuman preference · 2026-10-28 · rating

deepseek-llm-67b-chat

1,094.83 rating

Confidence interval: 1,076.34 – 1,113.33

1219 votes

Category
industry_software_and_it_services
Evaluated configuration
DeepSeek LLM 67B Chat
Evaluation type
human preference
Arena text preference / industry_writing_and_literature_and_languagehuman preference · 2026-10-28 · rating

deepseek-llm-67b-chat

1,110.60 rating

Confidence interval: 1,091.88 – 1,129.31

1142 votes

Category
industry_writing_and_literature_and_language
Evaluated configuration
DeepSeek LLM 67B Chat
Evaluation type
human preference
Arena text preference / instruction_followinghuman preference · 2026-10-28 · rating

deepseek-llm-67b-chat

1,078.93 rating

Confidence interval: 1,062.16 – 1,095.70

1525 votes

Category
instruction_following
Evaluated configuration
DeepSeek LLM 67B Chat
Evaluation type
human preference
Arena text preference / longer_queryhuman preference · 2026-10-28 · rating

deepseek-llm-67b-chat

1,091.69 rating

Confidence interval: 1,055.60 – 1,127.78

243 votes

Category
longer_query
Evaluated configuration
DeepSeek LLM 67B Chat
Evaluation type
human preference
Arena text preference / mathhuman preference · 2026-10-28 · rating

deepseek-llm-67b-chat

1,107.63 rating

Confidence interval: 1,084.26 – 1,131.01

576 votes

Category
math
Evaluated configuration
DeepSeek LLM 67B Chat
Evaluation type
human preference
Arena text preference / multi_turnhuman preference · 2026-10-28 · rating

deepseek-llm-67b-chat

1,081.94 rating

Confidence interval: 1,058.03 – 1,105.85

620 votes

Category
multi_turn
Evaluated configuration
DeepSeek LLM 67B Chat
Evaluation type
human preference
Arena text preference / non_englishhuman preference · 2026-10-28 · rating

deepseek-llm-67b-chat

1,072.75 rating

Confidence interval: 1,054.36 – 1,091.14

1112 votes

Category
non_english
Evaluated configuration
DeepSeek LLM 67B Chat
Evaluation type
human preference
Arena text preference / overallhuman preference · 2026-10-28 · rating

deepseek-llm-67b-chat

1,105.00 rating

Confidence interval: 1,093.48 – 1,116.53

4932 votes

Category
overall
Evaluated configuration
DeepSeek LLM 67B Chat
Evaluation type
human preference
Arena text preference / chinesehuman preference · 2026-10-15 · rating

deepseek-llm-67b-chat

1,131.80 rating

Confidence interval: 1,092.45 – 1,171.14

211 votes

Category
chinese
Evaluated configuration
deepseek-llm-67b-chat
Evaluation type
human preference
Arena text preference / codinghuman preference · 2026-10-15 · rating

deepseek-llm-67b-chat

1,095.38 rating

Confidence interval: 1,071.88 – 1,118.88

649 votes

Category
coding
Evaluated configuration
deepseek-llm-67b-chat
Evaluation type
human preference
Arena text preference / creative_writinghuman preference · 2026-10-15 · rating

deepseek-llm-67b-chat

1,067.27 rating

Confidence interval: 1,045.42 – 1,089.12

789 votes

Category
creative_writing
Evaluated configuration
deepseek-llm-67b-chat
Evaluation type
human preference
Arena text preference / englishhuman preference · 2026-10-15 · rating

deepseek-llm-67b-chat

1,127.01 rating

Confidence interval: 1,114.24 – 1,139.78

3820 votes

Category
english
Evaluated configuration
deepseek-llm-67b-chat
Evaluation type
human preference
Arena text preference / exclude_tieshuman preference · 2026-10-15 · rating

deepseek-llm-67b-chat

942.36 rating

Confidence interval: 925.69 – 959.03

3443 votes

Category
exclude_ties
Evaluated configuration
deepseek-llm-67b-chat
Evaluation type
human preference
Arena text preference / hard_promptshuman preference · 2026-10-15 · rating

deepseek-llm-67b-chat

1,069.67 rating

Confidence interval: 1,051.32 – 1,088.02

1120 votes

Category
hard_prompts
Evaluated configuration
deepseek-llm-67b-chat
Evaluation type
human preference
Arena text preference / hard_prompts_englishhuman preference · 2026-10-15 · rating

deepseek-llm-67b-chat

1,085.19 rating

Confidence interval: 1,065.33 – 1,105.05

946 votes

Category
hard_prompts_english
Evaluated configuration
deepseek-llm-67b-chat
Evaluation type
human preference
Arena text preference / industry_business_and_management_and_financial_operationshuman preference · 2026-10-15 · rating

deepseek-llm-67b-chat

1,047.62 rating

Confidence interval: 1,019.36 – 1,075.87

441 votes

Category
industry_business_and_management_and_financial_operations
Evaluated configuration
deepseek-llm-67b-chat
Evaluation type
human preference
Arena text preference / industry_entertainment_and_sports_and_mediahuman preference · 2026-10-15 · rating

deepseek-llm-67b-chat

1,061.67 rating

Confidence interval: 1,040.18 – 1,083.15

802 votes

Category
industry_entertainment_and_sports_and_media
Evaluated configuration
deepseek-llm-67b-chat
Evaluation type
human preference
Arena text preference / industry_legal_and_governmenthuman preference · 2026-10-15 · rating

deepseek-llm-67b-chat

1,089.04 rating

Confidence interval: 1,054.22 – 1,123.87

300 votes

Category
industry_legal_and_government
Evaluated configuration
deepseek-llm-67b-chat
Evaluation type
human preference
Arena text preference / industry_life_and_physical_and_social_sciencehuman preference · 2026-10-15 · rating

deepseek-llm-67b-chat

1,065.76 rating

Confidence interval: 1,043.98 – 1,087.55

876 votes

Category
industry_life_and_physical_and_social_science
Evaluated configuration
deepseek-llm-67b-chat
Evaluation type
human preference
Arena text preference / industry_mathematicalhuman preference · 2026-10-15 · rating

deepseek-llm-67b-chat

1,093.04 rating

Confidence interval: 1,066.51 – 1,119.56

484 votes

Category
industry_mathematical
Evaluated configuration
deepseek-llm-67b-chat
Evaluation type
human preference
Arena text preference / industry_medicine_and_healthcarehuman preference · 2026-10-15 · rating

deepseek-llm-67b-chat

1,061.61 rating

Confidence interval: 1,032.99 – 1,090.23

385 votes

Category
industry_medicine_and_healthcare
Evaluated configuration
deepseek-llm-67b-chat
Evaluation type
human preference
Arena text preference / industry_software_and_it_serviceshuman preference · 2026-10-15 · rating

deepseek-llm-67b-chat

1,094.92 rating

Confidence interval: 1,076.42 – 1,113.41

1219 votes

Category
industry_software_and_it_services
Evaluated configuration
deepseek-llm-67b-chat
Evaluation type
human preference
Arena text preference / industry_writing_and_literature_and_languagehuman preference · 2026-10-15 · rating

deepseek-llm-67b-chat

1,110.66 rating

Confidence interval: 1,091.94 – 1,129.38

1142 votes

Category
industry_writing_and_literature_and_language
Evaluated configuration
deepseek-llm-67b-chat
Evaluation type
human preference
Arena text preference / instruction_followinghuman preference · 2026-10-15 · rating

deepseek-llm-67b-chat

1,078.94 rating

Confidence interval: 1,062.16 – 1,095.71

1525 votes

Category
instruction_following
Evaluated configuration
deepseek-llm-67b-chat
Evaluation type
human preference
Arena text preference / longer_queryhuman preference · 2026-10-15 · rating

deepseek-llm-67b-chat

1,091.73 rating

Confidence interval: 1,055.61 – 1,127.85

243 votes

Category
longer_query
Evaluated configuration
deepseek-llm-67b-chat
Evaluation type
human preference
Arena text preference / mathhuman preference · 2026-10-15 · rating

deepseek-llm-67b-chat

1,107.59 rating

Confidence interval: 1,084.21 – 1,130.97

576 votes

Category
math
Evaluated configuration
deepseek-llm-67b-chat
Evaluation type
human preference
Arena text preference / multi_turnhuman preference · 2026-10-15 · rating

deepseek-llm-67b-chat

1,081.94 rating

Confidence interval: 1,058.02 – 1,105.86

620 votes

Category
multi_turn
Evaluated configuration
deepseek-llm-67b-chat
Evaluation type
human preference
Arena text preference / non_englishhuman preference · 2026-10-15 · rating

deepseek-llm-67b-chat

1,072.84 rating

Confidence interval: 1,054.45 – 1,091.24

1112 votes

Category
non_english
Evaluated configuration
deepseek-llm-67b-chat
Evaluation type
human preference
Arena text preference / overallhuman preference · 2026-10-15 · rating

deepseek-llm-67b-chat

1,104.99 rating

Confidence interval: 1,093.46 – 1,116.52

4932 votes

Category
overall
Evaluated configuration
deepseek-llm-67b-chat
Evaluation type
human preference

Run it locally

Inference · memory estimate

deepseek-llm-67b-chat

Not enough verified architecture information for a reliable VRAM estimate.

Capabilities

Context window
4,096 tokens
Open weights
Yes
Model license
other

Input modalities

  • text

Output modalities

  • text

Processing & platform

Architecture
llama

* Provisional data