PipelineScore
← Back to leaderboard
Head-to-head

StarCoder2 3B vs DeepSeek R1 671B-A37B

DeepSeek R1 671B-A37B wins+39.0 pts (+82%)on the balanced composite.
bigcode
StarCoder2 3B
47.3TAP
deepseek
Winner
DeepSeek R1 671B-A37B
86.3MAINLINE

Category-by-category

Higher bar wins each row.

StarCoder2 3B
Category
DeepSeek R1 671B-A37B
-39.847.7
Code
87.5+39.8
-36.646.0
Reason
82.7+36.6
-40.649.6
Tool Use
90.3+40.6
-32.550.4
RAG
82.9+32.5
-44.143.7
Speed
87.8+44.1
StarCoder2 3B
Provider
bigcode
Released
2024-02-28
Context
16.384K
Lab
Community
DeepSeek R1 671B-A37B
Provider
deepseek
Released
2025-01-20
Context
128K
Lab
Community
StarCoder2 3B vs DeepSeek R1 671B-A37B · PipelineScore