DeepSeek V2

DeepSeek · deepseek/deepseek-v2

Provider	DeepSeek
Family	deepseek
Type	llm-code
Status	active
Released	2024-04-22

Reported benchmark scores

These scores carry one collection date for the whole card (2026-04) and a source list rather than a source per score, so they cannot be attributed individually. They are shown as reported and are not verified evidence. Benchmark pages carry the reviewed, dated evidence where it exists.

Benchmark	Catalogue standing	Score
AlpacaEval	unassessed	42.8
Arena Elo — Coding	unassessed	1320
Arena Elo — Math	unassessed	1300
Arena Elo — Overall (Text)	unassessed	1310
BIG-Bench Hard	unassessed	64.1
BBQ (Bias Benchmark for QA)	unassessed	73.5
GPQA Diamond	unassessed	59.1
HELM Safety	unassessed	78.2
HumanEval	unassessed	85.8
IFEval	unassessed	84
MATH-500	unassessed	89.2
MMLU-Pro	unassessed	75.5
MT-Bench	unassessed	8.8
MuSR	unassessed	29.8
SWE-bench Verified	unassessed	42
ToxiGen	unassessed	82.5
WildBench	unassessed	70.2

Data

This card as JSON · Edit on GitHub