VECTOR WIREAI INTELLIGENCE
⌘K
PKT
Refresh Models —Deals —Regulatory —Sources —

Questions this page answers

Which is better, DeepSeek-V3.1-Base or Mistral 7B?
Across 12 shared benchmarks, DeepSeek-V3.1-Base scores higher on 12 and Mistral 7B on 0. The widest gap is MATH, where DeepSeek-V3.1-Base scores 62.6 against 13.1.

DeepSeek-V3.1-Base vs Mistral 7B

Across 12 shared benchmarks, DeepSeek-V3.1-Base scores higher on 12 and Mistral 7B on 0. The widest gap is MATH, where DeepSeek-V3.1-Base scores 62.6 against 13.1.

DeepSeekvsMistral12 shared benchmarks12–0 head-to-head
BenchmarkDeepSeek-V3.1-BaseMistral 7B
arc_challenge95.660
bbh88.256.1
C-Eval9047.4
GPQA Diamond5124.7
GSM8K91.452.2
hellaswag89.283.2
humaneval72.529.3
MATH62.613.1
MBPP74.651.1
mmlu87.464.2
MMLU-Pro58.830.9
winogrande85.978.4

Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing.