Qwen3-30B-A3B-Instruct-2507

Home/Models/Qwen/Qwen3-30B-A3B-Instruct-2507

Qwen3-30B-A3B-Instruct-2507

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

kind: original
quant: bf16 (16-bit)
family: qwen
modality: text-generation
root model: Qwen/Qwen3-30B-A3B-Instruct-2507
provider: Qwen
license: apache-2.0
language: —
context: 262,144
library: transformers
vRAM (GB): —
size on disk: —
downloads: 2,116,385
likes: 822
created: —
last modified: 2025-09-17T06:56:43+00:00 (2025)
huggingface: https://huggingface.co/Qwen/Qwen3-30B-A3B-Instruct-2507
tags: transformerssafetensorsqwen3_moetext-generationconversationalarxiv:2402.17463arxiv:2407.02490arxiv:2501.15383arxiv:2404.06654arxiv:2505.09388license:apache-2.0eval-resultsendpoints_compatibledeploy:azureregion:usdeploy:sagemaker

Benchmark scores (9)

benchmark	category	score	context	conf	evidence
MMLU	knowledge	90.4	MMLU-Redux	medium	`\| MMLU-Redux \| 90.4 \| 91.3 \| 90.6 \| 89.2 \| 84.1 \| 89.3 \|`
IFEval	instruction-following	82.3	IFEval	medium	`\| IFEval \| 82.3 \| 83.9 \| 84.3 \| 83.2 \| 83.7 \| 84.7 \|`
MMLU-Pro	knowledge	81.2	MMLU-Pro	medium	`\| MMLU-Pro \| 81.2 \| 79.8 \| 81.1 \| 75.2 \| 69.1 \| 78.4 \|`
MMLU-Pro	knowledge	75.8	MMLU-ProX	medium	`\| MMLU-ProX \| 75.8 \| 76.2 \| 78.3 \| 73.2 \| 65.1 \| 72.0 \|`
GPQA	reasoning	68.4	GPQA	medium	`\| GPQA \| 68.4 \| 66.9 \| 78.3 \| 62.9 \| 54.8 \| 70.4 \|`
BFCL (function calling)	agentic	64.7	BFCL-v3	medium	`\| BFCL-v3 \| 64.7 \| 66.5 \| 66.1 \| 68.0 \| 58.6 \| 65.1 \|`
AIME 2025	math	46.6	AIME25	medium	`\| AIME25 \| 46.6 \| 26.7 \| 61.6 \| 24.7 \| 21.6 \| 61.3 \|`
Chatbot Arena	chat	45.6	Arena-Hard v2*	medium	`\| Arena-Hard v2* \| 45.6 \| 61.9 \| 58.3 \| 52.0 \| 24.8 \| 69.0 \|`
LiveCodeBench	coding	45.2	LiveCodeBench v6 (25.02-25.05)	medium	`\| LiveCodeBench v6 (25.02-25.05) \| 45.2 \| 35.8 \| 40.1 \| 32.9 \| 29.0 \| 43.2 \|`

Discussions (32)

title	author	status	type
Qwen3 30B — what GPU setup do you need?	duola15	open	disc
OMS Score -- OMS 70.4 (B) -- Qwen3 30B A3B \| 5-dimension independent scoring	duola15	open	disc
TemporalMesh Transformer: 29.4 PPL at 48% compute — beats Mamba, new open-source architecture	vigneshwar234	open	disc
Regarding RoPE frequency	juneyongyang	open	disc
Add GPQA evaluation result	burtenshaw	open	disc
🦊 致 Qwen3：谢谢你成为“思考者”的基石	aifeifei798	open	disc
dummy config.json	michaelfeil	closed	disc
Lora Fine Tning- Several Issues	aetherforge	open	disc
RTX 5090 + Qwen 30B MoE @ 135 tok/s in NVFP4 Guide	JohnTdi	open	disc
Error when utilizing DUAL_CHUNK_FLASH_ATTN with vLLM.	lssj14	open	disc
Can I enable DCA on other qwen3 dense models?	dophys	open	disc
sarpelai	bsarpel	draft	disc
Accelerate + Megatron LM compatability	logan-vegna-shopify	closed	disc
Structured Output Not Supported in Qwen3-30B-A3B-Instruct-2507	navinnseab	closed	disc
Update README.md	Brokersponsor	open	disc
Polish language support	eXt73	open	disc
how to switch between Qwen3-30B-A3B-Instruct-2507 (Full Attention) and Qwen3-30B-A3B-Instruct-2507 (Sparse Attention)	Rocke518	closed	disc
Working GPTQ-Int4 version?	koesn	open	disc
fix: max tokens to 1M	heitormsb	open	disc
help: 1M context error?	heitormsb	open	disc
Add 1M support	hzhwcmhf	merged	disc
关于 static KV-Cache(ValueError: This model does not support cache_implementation='static')	luohuashijieyoufengjun	open	disc
Upload 17F9CDA2-7E35-4793-9D18-0581B0276089.jpeg	Faragcg	open	disc
How to achieve reliable native function calling	Ithanil	closed	disc
Test Scores Can Be Misleading	phil111	open	disc
Qwen3-30B-A3B-Instruct-2507-Autoround-Int-4bit-gptq	jart25	open	disc
Remove trailing tabs after license to avoid mlx_lm.convert errors	Gallardo994	open	disc
Local Installation Video and Testing - Step by Step	fahdmirzac	open	disc
Is the base model of the this model Qwen3-30B-A3B？	paulcx	open	disc
AWQ version	devops724	open	disc
An Improvement, But Q3 30b Still Has Very Little General Knowledge	phil111	open	disc
THANK YOU QWEN! 😍	AyyYOO	open	disc