Back to Benchmarks
GPT-4.1 mini vs DeepSeek R1 Distill Llama 70B
GPT-4.1 mini
OpenAI
vs
DeepSeek R1 Distill Llama 70B
DeepSeek
via DeepInfra
GPT-4.1 mini wins 6 of 11
Index Scores
22.9
+6.9
Intelligence
16.0
18.5
+7.1
Coding
11.4
46.3
Math
+7.4
53.7
Benchmarks
78.1
MMLU Pro
+1.4
79.5
66.4
+26.2
GPQA
40.2
4.6
HLE
+1.5
6.1
48.3
+21.7
LiveCode
26.6
40.4
+9.2
SciCode
31.2
92.5
MATH-500
+1.0
93.5
43.0
AIME
+24.0
67.0
Speed
51
+9
TPS
42
Shared
Winning delta
crafted by
bart stefanski