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 Qwen3 VL 4B (Reasoning)

MetavsAlibaba20 shared benchmarks127 head-to-head
BenchmarkLlama 3.1 Instruct 405BQwen3 VL 4B (Reasoning)
AA Agentic Index6.314.4
AA Intelligence8.57.7
AA-LCR24.723
AA-Omniscience-17.1-68.4
AIME 202423.372.9
Artificial Analysis Coding Index14.56.7
critpt00
gdpval013.8
GPQA Diamond51.564.1
HLE4.24.6
IFBench3936.6
ifeval88.682.6
LiveCodeBench30.151.3
mmlu_redux86.286
MMLU-Pro73.473.6
OmniScience Accuracy23.212.2
OmniScience Non-Hallucination498.2
scicode29.917.1
Terminal-Bench Hard6.81.5
τ²-Bench Telecom (AA run)1915.5

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.