HF Model & Benchmark Matrix

24,078 variants · 14,984 scores · 120 benchmarks
Home/Orgs/Qwen/Qwen/Qwen3-235B-A22B-Thinking-2507

Qwen/Qwen3-235B-A22B-Thinking-2507

family qwen · 14 variants · unknown

provider
Qwen
released
—
huggingface
https://huggingface.co/Qwen/Qwen3-235B-A22B-Thinking-2507
best scores
MMLU-Pro: 85.0MMLU: 93.4GPQA: 81.0AIME 2025: 87.5Humanity's Last Exam: 18.1LiveCodeBench: 68.7IFEval: 79.1Chatbot Arena: 72.2BFCL (function calling): 63.8MT-Bench: 1.83MATH: 96.9

Best scores per benchmark

benchmarkcategorybest score
MATHmath96.9
MMLUknowledge93.4
AIME 2025math87.5
MMLU-Proknowledge85.0
GPQAreasoning81.0
IFEvalinstruction-following79.1
Chatbot Arenachat72.2
LiveCodeBenchcoding68.7
BFCL (function calling)agentic63.8
Humanity's Last Examknowledge18.1
MT-Benchchat1.83

Variants (14)

variantkindquantbitsvRAMdlorgyear
Qwen3-235B-A22B-Thinking-2507-FP8fp8fp88.045,489Qwen2025
Qwen3-235B-A22B-Thinking-2507originalbf1616.020,387Qwen2025
Qwen3-235B-A22B-Thinking-2507-GGUFggufgguf16.04,546unsloth2025
Qwen3-235B-A22B-Thinking-2507-NVFP4nvfp4nvfp44.01,186nvidia2026
Qwen_Qwen3-235B-A22B-Thinking-2507-GGUFggufgguf1.01,145bartowski2025
Qwen3-Nemotron-235B-A22B-GenRM-2603originalbf1616.0799nvidia2026
Qwen3-235B-A22B-Thinking-2507-MLX-8bitmlxmlx750lmstudio-community2025
Qwen3-235B-A22B-Thinking-2507-MLX-4bitmlxmlx394lmstudio-community2025
Qwen3-235B-A22B-Thinking-2507-GGUFggufgguf3.0299lmstudio-community2025
Qwen3-235B-A22B-Thinking-2507-MLX-6bitmlxmlx276lmstudio-community2025
MiroThinker-1.7-mini-GGUFggufgguf16.0175Mungert2026
Qwen3-235B-A22B-Thinking-2507-Eagle3originalbf1616.0106nvidia2026
Qwen3-Nemotron-235B-A22B-GenRMoriginalbf1616.0100nvidia2025
AReaL-SEA-235B-A22Boriginalbf1616.031inclusionAI2026
Hugging Face model & benchmark data scraped & matrix-ified. 24,078 variants · 14,984 scores · 120 benchmarks.