Skip to content
Leaders Board.
Models/DeepSeekOpen weights

DeepSeek-R1 (May 2025)

Ranked on the boards below.
-
Overall rank
-
Podium score
0
Board wins
Jan 22, 2025
Released
·Chat·Reasoning·Coding

Chat

Which model people prefer in blind head-to-head conversations.

93
Text Arena
1,421 · of 100
as “deepseek-r1-0528”

Reasoning

Hard questions with checkable answers: science, math, puzzles, expert exams.

39
MMLU-Pro
84.0% · of 100 · stale since 2026-03-11
as “DeepSeek-R1”
60
ARC-AGI-2
1.3% · of 78
as “Deepseek R1”
81
GPQA Diamond
76.3% · of 100
as “DeepSeek-R1-0528”
99
Epoch ECI
141.3 · of 100
as “DeepSeek-R1 (May 2025)”

Coding

Fixing real repositories, building web apps, editing code.

12
Aider Polyglot
71.4% · of 50 · stale since 2025-10-03
as “DeepSeek R1 (0528)”

More from DeepSeek

DeepSeek V4.1 FlashDeepSeek V4 Pro 0813DeepSeek V4 Flash VisionDeepSeek-V3.2DeepSeekDeepSeek V3DeepSeek V3.1 TerminusDeepSeek V3.1

Scores as published by each board on the capture date. Model names and logos belong to their owners; logos via logo.dev.

Weekly: the AI leaderboards with a new #1, Saturday mornings.