HF Model & Benchmark Matrix

24,078 variants · 14,984 scores · 120 benchmarks
Home/Orgs/Qwen/Qwen/Qwen3-30B-A3B-Thinking-2507

Qwen/Qwen3-30B-A3B-Thinking-2507

family qwen · 13 variants · unknown

provider
Qwen
released
—
huggingface
https://huggingface.co/Qwen/Qwen3-30B-A3B-Thinking-2507
best scores
MMLU-Pro: 81.9MMLU: 92.1GPQA: 82.8AIME 2025: 72.0LiveCodeBench: 61.2IFEval: 89.8Chatbot Arena: 56.7BFCL (function calling): 68.6

Best scores per benchmark

benchmarkcategorybest score
MMLUknowledge92.1
IFEvalinstruction-following89.8
GPQAreasoning82.8
MMLU-Proknowledge81.9
AIME 2025math72.0
BFCL (function calling)agentic68.6
LiveCodeBenchcoding61.2
Chatbot Arenachat56.7

Variants (13)

variantkindquantbitsvRAMdlorgyear
Qwen3-30B-A3B-Thinking-2507originalbf1616.0141,325Qwen2025
Qwen3-30B-A3B-Thinking-2507-FP8fp8fp88.065,067Qwen2025
Qwen3-30B-A3B-Thinking-2507-GGUFggufgguf16.05,570unsloth2025
Qwen3-30B-A3B-Thinking-2507-GGUFggufgguf3.01,652lmstudio-community2025
Qwen_Qwen3-30B-A3B-Thinking-2507-GGUFggufgguf2.01,049bartowski2025
Qwen3-30B-A3B-Thinking-2507-MLX-8bitmlxmlx589lmstudio-community2025
Qwen3-30B-A3B-Thinking-2507-MLX-4bitmlxmlx548lmstudio-community2025
Qwen3-30B-A3B-Thinking-2507-GGUFggufgguf16.0542Mungert2025
Qwen3-30B-A3B-Thinking-2507-MLX-6bitmlxmlx479lmstudio-community2025
MiroThinker-v1.0-30B-GGUFggufgguf16.0369Mungert2025
Qwen3-30B-A3B-Thinking-2507originalbf1616.0293unsloth2025
MiroThinker-v1.5-30B-GGUFggufgguf1.0190Mungert2026
Qwen3-30B-A3B-Thinking-2507-Q8_0-GGUFggufgguf8.083ggml-org2025
Hugging Face model & benchmark data scraped & matrix-ified. 24,078 variants · 14,984 scores · 120 benchmarks.