DeepSeek V3 0324

DeepSeek · deepseek/deepseek-v3-0324

ProviderDeepSeek
Familydeepseek
Typellm-code
Statusactive
Released2025-03-24

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.
BenchmarkCatalogue standingScore
Aider Polyglot Benchmarkunassessed58.5
AlpacaEvalunassessed42.8
Arena Elo — Codingunassessed1330.0
Arena Elo — Hard Promptsunassessed1408.3
Arena Elo — Mathunassessed1310.0
Arena Elo — Overall (Text)unassessed1320.0
Arena Elo — Style Controlunassessed1394.6
Artificial Analysis Intelligence Indexunassessed80.0
Artificial Analysis Output Speedunassessed70.0
BIG-Bench Hardunassessed64.1
BBQ (Bias Benchmark for QA)unassessed73.5
GPQA Diamondunassessed59.1
GSM8Kunassessed89.3
HELM Safetyunassessed78.2
HumanEvalunassessed85.8
IFEvalunassessed84.0
LiveCodeBenchunassessed49.2
MATH-500unassessed89.2
MMLU-Prounassessed75.5
MT-Benchunassessed8.8
MultiPL-E: C#unassessed77.5
MultiPL-E: Juliaunassessed58.5
MultiPL-E: Kotlin (unconfirmed)unassessed69.2
MultiPL-E: Luaunassessed52.2
MultiPL-E: Perlunassessed47.5
MultiPL-E: PHPunassessed74.8
MultiPL-E: Runassessed54.2
MultiPL-E: Rubyunassessed62.1
MultiPL-E: Scalaunassessed57.2
MultiPL-E: Swiftunassessed62.5
MuSRunassessed29.8
SWE-bench Agentunassessed35.8
SWE-bench Verifiedunassessed42.0
τ-benchunassessed41.2
ToxiGenunassessed82.5
WildBenchunassessed70.2

Data

This card as JSON · Edit on GitHub