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

Llama 3.1 Instruct 405B vs o3

MetavsOpenAI31 shared benchmarks229 head-to-head
BenchmarkLlama 3.1 Instruct 405Bo3
AA Agentic Index6.336.1
AA Intelligence8.531.1
AA-LCR24.773.3
AA-Omniscience-17.1-15.3
aider_polyglot5.881.3
AIME 202423.391.6
AIR-Bench 202458.684.5
anthropic_red_team96.598.3
Artificial Analysis Coding Index14.538.4
bbq94.597.9
critpt01.1
Fortress20.616
gdpval012.8
GPQA Diamond51.583.3
harmbench62.798.4
HLE4.220.3
IFBench3971.4
LiveCodeBench30.184.7
longbench_v236.158.8
math_500_em73.898.1
MMLU-Pro73.485
OmniScience Accuracy23.238.6
OmniScience Non-Hallucination4912.9
scicode29.941
simple_safety_tests98.899
simplebench2353.1
simpleqa23.249.4
SWE-bench Verified24.569.1
Terminal-Bench Hard6.837.1
xstest95.997.3
τ²-Bench Telecom (AA run)1980.7

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.