VECTOR WIREAI INTELLIGENCE
NVDA$1,847+3.2%MSFT$512+1.1%GOOGL$199-0.4%META$728+2.7%AMD$184-1.2%TSM$212+0.6%PLTR$98+4.1%AI IDX4,821+1.9%
PKT
SEEDRefresh Models Deals Regulatory Sources

DeepSeek-V4-Flash vs Llama 3.1 Instruct 405B

DeepSeekvsMeta22 shared benchmarks211 head-to-head
BenchmarkDeepSeek-V4-FlashLlama 3.1 Instruct 405B
AA Agentic Index61.36.3
AA Intelligence51.88.5
AA-LCR74.324.7
AA-Omniscience-14.3-17.1
Artificial Analysis Coding Index69.114.5
Chinese SimpleQA (C-SimpleQA)78.954.7
Codeforces305225.3
critpt16.60
gdpval52.90
GPQA Diamond90.851.5
HLE38.64.2
IFBench79.239
LiveCodeBench91.630.1
MMLU-Pro86.473.4
OmniScience Accuracy40.423.2
OmniScience Non-Hallucination15.649
scicode49.929.9
simplebench46.323
SWE-bench Pro52.611.2
SWE-bench Verified7924.5
Terminal-Bench Hard38.66.8
τ²-Bench Telecom (AA run)95.619

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.