PipelineScore
← Back to leaderboard
Head-to-head

Qwen 2.5 1.5B Instruct vs DeepSeek V3 671B-A37B

DeepSeek V3 671B-A37B wins+38.4 pts (+88%)on the balanced composite.
deepseek
Winner
DeepSeek V3 671B-A37B
82.0MAINLINE

Category-by-category

Higher bar wins each row.

Qwen 2.5 1.5B Instruct
Category
DeepSeek V3 671B-A37B
-38.143.2
Code
81.3+38.1
-38.244.0
Reason
82.3+38.2
-35.249.0
Tool Use
84.3+35.2
-37.148.0
RAG
85.1+37.1
-33.746.0
Speed
79.7+33.7
Qwen 2.5 1.5B Instruct
Provider
alibaba
Released
2024-09-19
Context
32.768K
Lab
Community
DeepSeek V3 671B-A37B
Provider
deepseek
Released
2024-12-26
Context
128K
Lab
Community