Which is better, Llama 3.1 Instruct 405B or Qwen3.5 4B?

Llama 3.1 Instruct 405B vs Qwen3.5 4B

22 SHARED BENCHMARKS

Across 22 shared benchmarks, Llama 3.1 Instruct 405B scores higher on 5 and Qwen3.5 4B on 16, with 1 level. The widest gap is AA Agentic Index, where Qwen3.5 4B scores 36.3 against 6.3.

METAVSALIBABA22 SHARED5–16 HEAD-TO-HEAD

BenchmarkLlama 3.1 Instruct 405BMARGINQwen3.5 4B

AA Agentic Index6.336.3

AA Intelligence8.520

AA-LCR25.363

AA-Omniscience-17.1-58.4

Artificial Analysis Coding Index14.522.6

bfcl_v488.554

C-Eval72.585.1

critpt00

gdpval08.3

GPQA Diamond51.577.1

HLE4.29.9

IFBench3959.2

ifeval88.689.8

longbench_v236.150

mmlu_redux86.288.8

MMLU-Pro73.479.1

mmmlu73.876.1

OmniScience Accuracy23.215.1

OmniScience Non-Hallucination4913.4

scicode29.918.3

Terminal-Bench Hard6.818.2

τ²-Bench Telecom (AA run)1992.1

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.

Compare Llama 3.1 Instruct 405B withALL PAIRINGS →

MetaMuse Spark 1.1

Across the fieldGemini 3.6 Flash Nemotron 3 Super 120B A12B GPT-5.5 Claude Opus 4.8 GPT-4o GPT-5.6 Sol Gemini 3.1 Pro GPT-5.4

Compare Qwen3.5 4B withALL PAIRINGS →

AlibabaQwen3.5 35B A3B Qwen3 VL 4B (Reasoning)Qwen3.5 397B A17B Qwen2.5 Instruct 72B Qwen3.5 27B Qwen3.6 Plus

Across the fieldDeepSeek-V3.2 DeepSeek-V4-Flash Claude Opus 4.7 Gemini 2.5 Pro Gemini 1.5 Pro Claude Sonnet 4 Claude Sonnet 5 Muse Spark 1.1