Qwen3-4B-Thinking-2507

Home/Models/Qwen/Qwen3-4B-Thinking-2507

Qwen/Qwen3-4B-Thinking-2507

kind: original
quant: bf16 (16-bit)
family: qwen
modality: text-generation
root model: Qwen/Qwen3-4B-Thinking-2507
provider: Qwen
license: apache-2.0
language: —
context: 262,144
library: transformers
vRAM (GB): —
size on disk: —
downloads: 294,867
likes: 606
created: —
last modified: 2025-08-06T11:08:25+00:00 (2025)
huggingface: https://huggingface.co/Qwen/Qwen3-4B-Thinking-2507
tags: transformerssafetensorsqwen3text-generationconversationalarxiv:2505.09388license:apache-2.0eval-resultstext-generation-inferenceendpoints_compatibledeploy:azureregion:us

benchmark	category	score	context	conf	evidence
MMLU	knowledge	89.5	MMLU-Redux	medium	`\| MMLU-Redux \| 89.5 \| 83.7 \| 86.1 \|`
IFEval	instruction-following	86.5	IFEval	medium	`\| IFEval \| 86.5 \| 81.9 \| 87.4 \|`
MMLU-Pro	knowledge	78.5	MMLU-Pro	medium	`\| MMLU-Pro \| 78.5 \| 70.4 \| 74.0 \|`
MMLU-Pro	knowledge	73.1	MMLU-ProX	medium	`\| MMLU-ProX \| 73.1 \| 61.0 \| 64.2 \|`
AIME 2025	math	70.9	AIME25	medium	`\| AIME25 \| 70.9 \| 65.6 \| 81.3 \|`
BFCL (function calling)	agentic	69.1	BFCL-v3	medium	`\| BFCL-v3 \| 69.1 \| 65.9 \| 71.2 \|`
GPQA	reasoning	65.8	GPQA	medium	`\| GPQA \| 65.8 \| 55.9 \| 65.8 \|`
LiveCodeBench	coding	57.4	LiveCodeBench v6 (25.02-25.05)	medium	`\| LiveCodeBench v6 (25.02-25.05) \| 57.4 \| 48.4 \| 55.2 \|`
Chatbot Arena	chat	36.3	Arena-Hard v2$	medium	`\| Arena-Hard v2$ \| 36.3 \| 13.7 \| 34.9 \|`

none scraped.