HF Model & Benchmark Matrix

24,078 variants · 14,984 scores · 120 benchmarks
Home/Orgs/Qwen/Qwen/Qwen3-Next-80B-A3B-Thinking

Qwen/Qwen3-Next-80B-A3B-Thinking

family qwen · 12 variants · unknown

provider
Qwen
released
—
huggingface
https://huggingface.co/Qwen/Qwen3-Next-80B-A3B-Thinking
best scores
MMLU-Pro: 80.9MMLU: 91.4GPQA: 73.4AIME 2025: 85.0LiveCodeBench: 66.0IFEval: 88.9Chatbot Arena: 56.0BFCL (function calling): 72.4

Best scores per benchmark

benchmarkcategorybest score
MMLUknowledge91.4
IFEvalinstruction-following88.9
AIME 2025math85.0
MMLU-Proknowledge80.9
GPQAreasoning73.4
BFCL (function calling)agentic72.4
LiveCodeBenchcoding66.0
Chatbot Arenachat56.0

Variants (12)

variantkindquantbitsvRAMdlorgyear
Qwen3-Next-80B-A3B-Thinkingoriginalbf1616.035,946Qwen2025
Qwen3-Next-80B-A3B-Thinking-FP8fp8fp88.024,453Qwen2025
Qwen_Qwen3-Next-80B-A3B-Thinking-GGUFggufgguf1.09,615bartowski2026
Qwen3-Next-80B-A3B-Thinking-GGUFggufgguf16.09,176unsloth2026
Qwen3-Next-80B-A3B-Thinking-GGUFggufgguf4.03,184Qwen2025
Qwen3-Next-80B-A3B-Thinking-NVFP4nvfp4nvfp44.01,912nvidia2026
Qwen3-Next-80B-A3B-Thinking-GGUFggufgguf4.01,565lmstudio-community2026
Qwen3-Next-80B-A3B-Thinking-MLX-4bitmlxmlx322lmstudio-community2025
Qwen3-Next-80B-A3B-Thinking-MLX-8bitmlxmlx206lmstudio-community2025
Qwen3-Next-80B-A3B-Thinking-MLX-6bitmlxmlx147lmstudio-community2025
Qwen3-Next-80B-A3B-Thinking-MLX-5bitmlxmlx132lmstudio-community2025
Qwen3-Next-80B-A3B-Thinkingoriginalbf1616.097unsloth2025
Hugging Face model & benchmark data scraped & matrix-ified. 24,078 variants · 14,984 scores · 120 benchmarks.