Qwen3-30B-A3B-Instruct-2507

30.5B parameters, 3B active (8 of 128 experts per token), text in, text out, 256K tokens of context. 4 providers serve it from $0.090 per million input tokens.

Verified 2026-10-25 Rank #83 in LLMs · score 102.8 method v1 Sources 4 Downloads 30d 1.1M · #105 of 619 tracked

Changed this week: 3 new quantized builds on the Hub (141 in all).2026-10-24

Can I run it?18GB at INT4Estimatefits a 24 GB cardmethodCommunity: Run Qwen3-30B-A3B-Instruct-2507 Locally | Atomic Chat →

What does it cost?$0.090/1M incheapest of 3DekaLLM2026-10-24

Can I use it?Apache 2.0Permissive; patent grantlicence file2026-10-24

Does it work?41.39on BFCL#10 of 28 measuredleaderboard2026-10-25

Do people use it?1.1Mdownloads, 30 days#105 of 619 we track · top 17%+3 likes, downloads +2% this week850 likesHugging Face2026-10-24

Worth reading

Picked by rule from 19 tracked pieces: the lab's own post, the most shared, a field report, the latest.

Most discussed · 490 HN pts

GPT-OSS vs. Qwen3 and a detailed look how things evolved since GPT-2

OpenAI released gpt-oss-120b and gpt-oss-20b, their first open-weight models since GPT-2 in 2019.

A 2025 research paper confirms Dropout results in worse downstream performance in single-epoch regimes.

magazine.sebastianraschka.com 2025-08-10 490 HN pts · 97 comments 5.2k words about the Qwen3 family family

Latest · 8 months ago

Qwen3.5 122B and 35B models offer Sonnet 4.5 performance on local computers

venturebeat.com 2026-02-28 461 HN pts · 272 comments about the Qwen3 family family

19 best of 19 tracked

The 3 picks above, plus these 16. Top is the radar's score: discussion, source authority and the lab's own posts, fading with age.

ArticleQwen3-Coder-Nextqwen.ai · 2026-02-03 · official
ArticleQwen3-Max-Thinkingqwen.ai · 2026-01-26 · official
ArticleQwen3-TTS family is now open sourced: Voice design, clone, and generationqwen.ai · 2026-01-22 · official
ArticleQwen3: Think deeper, act fasterqwenlm.github.io · 2025-04-28 · 869 HN pts · official
ArticleQwen3-Coder: Agentic coding in the worldqwenlm.github.io · 2025-07-22 · official
ArticleQwen3-Nextqwen.ai · 2025-09-12 · official
ArticleQwen3-VLqwen.ai · 2025-09-23 · official
CodeQwen3-Omni: Native Omni AI model for text, image and videogithub.com · 2025-09-22
ArticleA 30B Qwen model walks into a Raspberry Pi and runs in real timebyteshape.com · 2026-01-06 · 363 HN pts
ArticleQwen3 30B A3B 2507 Instruct Intelligence, Performance & ...artificialanalysis.ai · undated
ArticleQwen3-30B-A3B-Instruct-2507: Efficient Open-Source MoE LLM with 262K Context | DEV.codev.co · undated
ArticleQwen/Qwen3-30B-A3B-Instruct-2507 · An Improvement ...huggingface.co · undated
ArticleQwen: Qwen3 30B A3B Instruct 2507 - AI Model Details & Be...benchable.ai · undated
ArticleQwen3 30B A3B Instruct 2507 - API, Specs, Playground & Pricing - Puter Developerdeveloper.puter.com · undated
ArticleSwitching to Qwen3-30B-A3B-Instruct-2507 Significantly Improves Local Agenthexeditreality.com · 2025-07-31
ArticleQwen: Qwen3 30B A3B Instruct 2507 Review | Pricing, Benchmarks & Capabilities (2026) | Design for Onlinedesignforonline.com · 2026-02-25

Show all 16

Where to run it

Live hosted endpoints, cheapest first, USD per million tokens. Ties break on output price, then provider name.

3 providers from $0.090

3 providers with a live price, from $0.090 to $0.10 per million input tokens; the median is $0.090. One dot per provider, cheapest in teal.

#	Cheapest 4	In /1M	Out /1M	Context	Quant	Seen
1	DekaLLM	$0.090	$0.30	256K	unknown	Oct 24
2	SiliconFlow	$0.090	$0.30	256K	fp8	Oct 24
3	Nebius	$0.10	$0.30	256K	fp8	Oct 24
4	Featherless	—	—	—	—	Oct 25

All 4 providers →

Where it's used

Hugging Face does not say who downloads a model or with what. These are the proxies: API providers listing it, quantized builds, fine-tunes/adapters/merges, and Spaces that load it.

Served by 4 API providers; heavily run locally (141 quantized builds); a big base for builders (261 fine-tunes, adapters and merges).

As a service4DekaLLM, Featherless, Nebius, SiliconFlow
On own hardware141quantized builds
Built on261121 fine-tunes · 128 adapters · 12 merges
In apps82Hugging Face Spaces

Top Spaces running it: Alibaba-NLP/Tongyi-DeepResearch ♥ 64 · hfmlsoc/different-flops ♥ 11 · milwright/chatui-helper ♥ 10 · milwright/chat-adventure-games ♥ 3

Will it fit?

Weights only, an estimate: leave room for the KV cache. Method

Fits a 64 GB Mac (unified memory) at INT4

About 18 GB at INT4, 37 GB at FP8, 73 GB at BF16: weights only, so leave room for the KV cache. Every expert must be resident, so total parameters count.

Estimate params × bytes per param × 1.2 · how we estimate

How it scores

Third-party measurements only; author-reported numbers stay in Reference, marked as such.

#10 of 28 on BFCL so far

41.39. Each tick is a model on this site with a measured BFCL row; author-reported numbers stay in Reference. Percent (overall accuracy). FC

Source · leaderboard · 2026-10-25

The story so far

Weights, every dated article, and repo commits on one rail.

Most coverage came in the first two weeks

Teal is the weights (2025-07-28), amber triangles are 4 dated articles (each links to it), grey ticks are repo commits.

What people built on it

402 derivatives on the Hub. Which build to run, by runtime: the most downloaded in each class.

402 derivatives on the Hub, led by 141 quantized

From the Hugging Face model tree; quantized builds in teal, fine-tunes, merges and adapters in greys.

GGUF unsloth/Qwen3-30B-A3B-Instruct-2507-GGUF for llama.cpp, Ollama, LM Studio 3.5M downloads MLX lmstudio-community/Qwen3-30B-A3B-Instruct-2507-MLX-4bit for Apple silicon 907.2k downloads FP8 Qwen/Qwen3-30B-A3B-Instruct-2507-FP8 for vLLM, SGLang 5.8M downloads

All variants →

Reference

Everything else, folded: the full tables, every source.

Benchmarks

16 rows

Benchmarks 16 rows

measured third-party leaderboard row (2)author-reported number from the model card or eval files (14)No confidence intervals unless the source gave them; settings differ between rows, so compare within a benchmark, not across. Date = the benchmark's own date when it gives one, else when we read it.

Benchmark	Value	Setting / notes	Reported by	Date
measured BFCL · Call tools and APIs	41.39 percent (overall accuracy)	FC 5 sub-scores Non-Live AST Acc: 85.77 Live Acc: 77.94 Multi Turn Acc: 30 Web Search Acc: 22.5 Memory Acc: 17.63	leaderboard	2026-10-25
measured BFCL · Call tools and APIs	36.7 percent (overall accuracy)	Prompt 5 sub-scores Non-Live AST Acc: 88.92 Live Acc: 78.39 Multi Turn Acc: 23.5 Web Search Acc: 17.5 Memory Acc: 9.68	leaderboard	2026-10-25
author-reported AIME 2025 · Reason through hard problems	61.3	—	model author	2026-10-07
author-reported Aider Polyglot · Write code with agents	35.6	—	model author	2026-10-07
author-reported Arena-Hard · Chat and follow instructions	69	—	model author	2026-10-07
author-reported Creative Writing · Chat and follow instructions	86	—	model author	2026-10-07
author-reported GPQA Diamond · Reason through hard problems	70.4	—	model author	2026-10-07
author-reported HMMT · Reason through hard problems	43	—	model author	2026-10-07
author-reported IFEval · Chat and follow instructions	84.7	—	model author	2026-10-07
author-reported LiveBench · Reason through hard problems	69	—	model author	2026-10-07
author-reported LiveCodeBench · Write code with agents	43.2	—	model author	2026-10-07
author-reported MMLU · Reason through hard problems	89.3	—	model author	2026-10-07
author-reported MMLU-Pro · Reason through hard problems	78.4	—	model author	2026-10-07
author-reported MultiPL-E · Write code with agents	83.8	—	model author	2026-10-07
author-reported SuperGPQA · Reason through hard problems	53.4	—	model author	2026-10-07
author-reported ZebraLogic	90	—	model author	2026-10-07

What the card says

1 cited sentence

Summary generated, cited

It supports a native context length of 262,144 and is released under the Apache 2.0 license.^[1]^[2]

“262,144 natively”
license

Generated 2026-10-07 15:01 UTC by parasail-gemma-4-26b-a4b-it; each numeral points at the fact it rests on. Not a verdict.

Use-case evidence

6 use cases

What it is good at 6 use cases

Labels are evidence, not verdicts: measured a third-party leaderboard row for this model; author-reported a number the model author published in the card or eval files; structural a deterministic signal from the name, config or chat template, or a decision-model estimate. Each chip links to its source. All use cases.

Facts

24 fields

Listed in LLMs · MoE

Facts

Qwen3-30B-A3B-Instruct-2507 has 30.5B parameters (3B active per token). (safetensors metadata, observed 2026-10-24)
Qwen3-30B-A3B-Instruct-2507 is a mixture-of-experts model with 128 experts, routing 8 per token. (config.json, observed 2026-10-24)
Context length is 262,144 tokens (256K). (config.json max_position_embeddings, observed 2026-10-24)
License: Apache 2.0 (friction: none). Permissive; patent grant. (license file, observed 2026-10-24)
Created 2025-07-28, last modified 2025-09-17. (HF API, observed 2026-10-24)

Pipeline tag	text-generation
Library	transformers
Architectures	Qwen3MoeForCausalLM
Model type	qwen3_moe
Dtype	bfloat16
Params by dtype	BF16: 30,532,122,624
Size band	L
MoE	yes
Experts	128
Experts per token	8
Context length	262,144 (config.json max_position_embeddings)
Papers	arXiv:2402.17463, arXiv:2407.02490, arXiv:2501.15383, arXiv:2404.06654, arXiv:2505.09388
License	Apache 2.0 (apache-2.0)
Hub snapshot	2026-10-25

Servability

vLLM · SGLang

Servability

vLLM: supported · Qwen3MoeForCausalLM · 7050ef881ff0835f6890a47db1859a24c65481d6SGLang: supported · Qwen3MoeForCausalLM · 773cb938fe9484a8f7e7e46fb1b1fe96412c59ed

vLLM registry (7050ef881ff0835f6890a47db1859a24c65481d6) · SGLang registry (773cb938fe9484a8f7e7e46fb1b1fe96412c59ed) · Matched by architecture class name against each project's registry on main; 'supported' means the architecture is registered, not that this checkpoint was tested.

Hardware fit

3 estimates

Hardware fit ESTIMATE

Weights-only memory for Qwen3-30B-A3B-Instruct-2507: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.

BF16 (2 bytes/param)	73 GB · Single high-end GPU (<80 GB)
FP8 (1 byte/param)	37 GB · Single high-end GPU (<80 GB)
INT4 (0.5 bytes/param)	18 GB · Single GPU (<24 GB)

Weights only (params × bytes/param × 1.2 overhead); KV cache and activations extra. MoE: all experts must be resident, so total parameters count.

All variants

40 variants

Lineage

Original model Family: Qwen/Qwen3 402 derivatives on the Hub

quantized: 141finetune: 121merge: 12adapter: 128

Top variants (40 shown, ranked by downloads)

Variant	Relation	Quant	Params	Band	License	Author	Likes	Downloads
Qwen/Qwen3-30B-A3B-Instruct-2507-FP8	quantized	-FP8	30.5B	L	apache-2.0	Qwen	143	5.8M
unsloth/Qwen3-30B-A3B-Instruct-2507-GGUF	quantized	GGUF	—	—	apache-2.0	unsloth	324	3.5M
cyankiwi/Qwen3-30B-A3B-Instruct-2507-AWQ-4bit	quantized	AWQ	5.31B	M	apache-2.0	cyankiwi	35	3.3M
bineric/lynx-instruct-30b	finetune	—	30.5B	L	apache-2.0	bineric	5	1.2M
MaziyarPanahi/Qwen3-30B-A3B-Instruct-2507-GGUF	quantized	GGUF	—	—	—	MaziyarPanahi	4	1M
lmstudio-community/Qwen3-30B-A3B-Instruct-2507-MLX-4bit	quantized	MLX	30.5B	L	apache-2.0	lmstudio-community	8	907.2k
lmstudio-community/Qwen3-30B-A3B-Instruct-2507-MLX-8bit	quantized	MLX	30.5B	L	apache-2.0	lmstudio-community	3	873.2k
stelterlab/Qwen3-30B-A3B-Instruct-2507-AWQ	quantized	AWQ	30.8B	L	apache-2.0	stelterlab	1	870.9k
lmstudio-community/Qwen3-30B-A3B-Instruct-2507-MLX-6bit	quantized	MLX	30.5B	L	apache-2.0	lmstudio-community	0	860k
mlx-community/Qwen3-30B-A3B-Instruct-2507-4bit	quantized	-4bit	30.5B	L	apache-2.0	mlx-community	16	353k
lmstudio-community/Qwen3-30B-A3B-Instruct-2507-GGUF	quantized	GGUF	—	—	apache-2.0	lmstudio-community	11	314.9k
Applied-Innovation-Center/Karnak-40B-v1.0	finetune	—	40.7B	XL	apache-2.0	Applied-Innovation-Center	73	250.4k
QuantTrio/Qwen3-30B-A3B-Instruct-2507-GPTQ-Int8	quantized	GPTQ	30.5B	L	apache-2.0	QuantTrio	9	228.8k
bartowski/Qwen_Qwen3-30B-A3B-Instruct-2507-GGUF	quantized	GGUF	—	—	—	bartowski	30	211.1k
unsloth/Qwen3-30B-A3B-Instruct-2507	finetune	—	30.5B	L	apache-2.0	unsloth	13	119.3k
NVFP4/Qwen3-30B-A3B-Instruct-2507-FP4	quantized	—	15.6B	L	apache-2.0	NVFP4	12	101.4k
DevQuasar/Qwen.Qwen3-30B-A3B-Instruct-2507-W4A16-GPTQ	quantized	-W4A16	4.61B	M	—	DevQuasar	0	100.3k
RedHatAI/Qwen3-30B-A3B-Instruct-2507-quantized.w4a16	quantized	-quant	4.59B	M	apache-2.0	RedHatAI	3	68.5k
JunHowie/Qwen3-30B-A3B-Instruct-2507-GPTQ-Int4	quantized	GPTQ	30.5B	L	apache-2.0	JunHowie	0	66.4k
huihui-ai/Huihui-Qwen3-30B-A3B-Instruct-2507-abliterated	finetune	—	30.5B	L	apache-2.0	huihui-ai	41	62.8k
byteshape/Qwen3-30B-A3B-Instruct-2507-GGUF	quantized	GGUF	—	—	apache-2.0	byteshape	84	49.7k
cyankiwi/Qwen3-30B-A3B-Instruct-2507-AWQ-8bit	quantized	AWQ	9.04B	L	apache-2.0	cyankiwi	2	51.8k
RedHatAI/Qwen3-30B-A3B-Instruct-2507-quantized.w8a8	quantized	-quant	30.6B	L	apache-2.0	RedHatAI	0	33.1k
ELVISIO/Qwen3-30B-A3B-Instruct-2507-AWQ	quantized	AWQ	30.5B	L	apache-2.0	ELVISIO	1	30.6k
pearl-ai/Qwen3-30B-A3B-Instruct-2507-pearl	quantized	—	30.5B	L	apache-2.0	pearl-ai	0	29.6k
INCModel/Qwen3-30B-A3B-Instruct-2507-MXFP4-CT-AutoRound	quantized	MXFP4	16.9B	L	apache-2.0	INCModel	0	26.5k
Intel/Qwen3-30B-A3B-Instruct-2507-int4-AutoRound	quantized	int4	635M	S	apache-2.0	Intel	7	21.2k
mradermacher/Qwen3-30B-A3B-Instruct-2507-i1-GGUF	quantized	GGUF	—	—	apache-2.0	mradermacher	3	14k
ramblingpolymath/Qwen3-30B-A3B-Instruct-2507-W4A16	quantized	-W4A16	30.8B	L	apache-2.0	ramblingpolymath	0	10.4k
ubergarm/Qwen3-30B-A3B-Instruct-2507-GGUF	quantized	GGUF	—	—	apache-2.0	ubergarm	14	8.5k
Text2MotionPrompter/Text2MotionPrompter	finetune	—	30.5B	L	apache-2.0	Text2MotionPrompter	15	7.2k
ramblingpolymath/Qwen3-30B-A3B-Instruct-2507-W8A8	quantized	-W8A8	30.6B	L	apache-2.0	ramblingpolymath	1	7.6k
perplexity-ai/browsesafe	finetune	—	30.5B	L	mit	perplexity-ai	52	4.5k
second-state/Qwen3-30B-A3B-Instruct-2507-GGUF	quantized	GGUF	—	—	apache-2.0	second-state	1	6.3k
OpenMOSS-Team/SciJudge-30B	finetune	—	30.5B	L	apache-2.0	OpenMOSS-Team	13	5.3k
apple/SimpleSD-30B-instruct	finetune	—	30.5B	L	apple-amlr	apple	11	5.3k
ggml-org/Qwen3-30B-A3B-Instruct-2507-Q8_0-GGUF	quantized	Q8_0	—	—	apache-2.0	ggml-org	1	5.7k
Orion-zhen/Qwen3-30B-A3B-Instruct-2507-Q4_K_M-GGUF	quantized	GGUF	—	—	—	Orion-zhen	2	5.6k
Intel/Qwen3-30B-A3B-Instruct-2507-gguf-q2ks-mixed-AutoRound	quantized	gguf	—	—	apache-2.0	Intel	25	4.3k
INCModel/Qwen3-30B-A3B-Instruct-2507-MXFP8-CT-AutoRound	quantized	FP8-	31.5B	L	apache-2.0	INCModel	0	5.6k

Repository changelog

21 commits

changed in the last 7 daysOther: 2026-10-24all changes

Repository changelog 21 commits

Date	Commit	SHA
2025-09-17	Update tokenizer_config.json	0d7cf23
2025-08-17	Update README.md	61082d4
2025-08-08	Add 1M support (#12)	3ffd1f5
2025-07-30	Update README.md	3d729a0
2025-07-29	Update LICENSE	01ee60e
2025-07-29	Update generation_config.json	97f3f7d
2025-07-29	Create LICENSE	c950284
2025-07-29	Update tokenizer_config.json	fd5492e
2025-07-29	Update README.md	bf85985
2025-07-29	Update README.md	b01caad
2025-07-29	Update README.md	6af341f
2025-07-29	Update generation_config.json	023ad30
2025-07-29	Update README.md	13b91ea
2025-07-29	Update README.md	014ba21
2025-07-28	Upload tokenizer.json	506b596
2025-07-28	Delete tokenizer.json	ee31d07
2025-07-28	Create generation_config.json	b92fe55
2025-07-28	Create README.md	9729168
2025-07-28	Update config.json	e67ac5d
2025-07-28	Upload folder using huggingface_hub	4457916
2025-07-28	initial commit	c9051e5

All providers

4 providers · 4 endpoints

Where to run 4 providers · 4 endpoints

Hosted endpoints observed for Qwen3-30B-A3B-Instruct-2507, sorted by input price (USD per million tokens). Each row names its observer and date; prices change daily, so check the source before quoting. Cheapest observed input price: $0.09/M.

Provider	Model ref	Context	Quant	$/M in	$/M out	tok/s	TTFT	Tools	Status	Observed
DekaLLM openrouter	qwen/qwen3-30b-a3b-instruct-2507	256K	unknown	$0.09	$0.30	—	—	yes	live	2026-10-24
SiliconFlow openrouter	qwen/qwen3-30b-a3b-instruct-2507	256K	fp8	$0.09	$0.30	—	—	yes	live	2026-10-24
Nebius openrouter	qwen/qwen3-30b-a3b-instruct-2507	256K	fp8	$0.10	$0.30	—	—	yes	live	2026-10-24
Featherless hf	Qwen/Qwen3-30B-A3B-Instruct-2507	—	—	—	—	—	—	—	live	2026-10-25

Articles and coverage

19 pieces

Ranked by source signals (HN, Lobsters, authority domains, official posts) and recency. Summaries are generated from a stored copy of the article text; every sentence cites a span of that text, and nothing is shown that failed the citation check.

Official1

Qwen3: Think deeper, act faster

qwenlm.github.io2025-04-28officialauthorityHN front pageabout the Qwen3 familyHN 869 pts · 388 comments2,139 words

Qwen3 introduces Hybrid Thinking Modes, allowing users to toggle between step-by-step reasoning and rapid responses via computational budget control. The open-weight MoE model Qwen3-235B-A22B achieves performance similar to dense models while using only 10% of the active parameters. Pre-training utilized approximately 36 trillion tokens across 119 languages, with deployment supported by SGLang and vLLM.written, reviewed, cited

Qwen3-8B and Qwen3-14B models do not use tied embeddings, unlike smaller variants.
The pre-training pipeline includes a final stage extending context length to 32K tokens.

Analysis9

Qwen3.5 122B and 35B models offer Sonnet 4.5 performance on local computers

venturebeat.com2026-02-28HN front pageabout the Qwen3 familyHN 461 pts · 272 comments

GPT-OSS vs. Qwen3 and a detailed look how things evolved since GPT-2

magazine.sebastianraschka.com2025-08-10authorityHN front pageabout the Qwen3 familyHN 490 pts · 97 comments5,195 words

OpenAI released gpt-oss-120b and gpt-oss-20b, their first open-weight models since GPT-2 in 2019. The 20B model can run on a consumer GPU with up to 16 GB of RAM, and the 120B model can run on a single H100 with 80 GB of RAM or newer hardware. The models utilize MXFP4 optimization to fit onto single GPUs.written, reviewed, cited

A 2025 research paper confirms Dropout results in worse downstream performance in single-epoch regimes.

A 30B Qwen model walks into a Raspberry Pi and runs in real time

byteshape.com2026-01-06HN front pageHN 363 pts · 135 comments3,728 words

Qwen3 30B A3B 2507 Instruct Intelligence, Performance & ...

artificialanalysis.aiauthority

Qwen3-30B-A3B-Instruct-2507: Efficient Open-Source MoE LLM with 262K Context | DEV.co

dev.co

Qwen/Qwen3-30B-A3B-Instruct-2507 · An Improvement ...

huggingface.co

Qwen: Qwen3 30B A3B Instruct 2507 - AI Model Details & Be...

benchable.ai

Run Qwen3-30B-A3B-Instruct-2507 Locally | Atomic Chat

atomic.chat

Qwen3 30B A3B Instruct 2507 - API, Specs, Playground & Pricing - Puter Developer

developer.puter.com

All coverage tracked for this model (19) — append-only, nothing is ever removed

Title	Source	Published	First seen	Signals
Qwen3 30B A3B 2507 Instruct Intelligence, Performance & ...	artificialanalysis.ai	—	2026-10-07	exa
Qwen3-30B-A3B-Instruct-2507: Efficient Open-Source MoE LLM with 262K Context \| DEV.co	dev.co	—	2026-10-07	exa
Qwen/Qwen3-30B-A3B-Instruct-2507 · An Improvement ...	huggingface.co	—	2026-10-07	exa
Qwen: Qwen3 30B A3B Instruct 2507 - AI Model Details & Be...	benchable.ai	—	2026-10-07	exa
Run Qwen3-30B-A3B-Instruct-2507 Locally \| Atomic Chat	atomic.chat	—	2026-10-07	exa
Qwen3 30B A3B Instruct 2507 - API, Specs, Playground & Pricing - Puter Developer	developer.puter.com	—	2026-10-07	exa
Qwen3.5 122B and 35B models offer Sonnet 4.5 performance on local computers	venturebeat.com	2026-02-28	2026-10-07	hn
Qwen: Qwen3 30B A3B Instruct 2507 Review \| Pricing, Benchmarks & Capabilities (2026) \| Design for Online	designforonline.com	2026-02-25	2026-10-07	exa
Qwen3-Coder-Next	qwen.ai official	2026-02-03	2026-10-07	hn
Qwen3-Max-Thinking	qwen.ai official	2026-01-26	2026-10-07	hn
Qwen3-TTS family is now open sourced: Voice design, clone, and generation	qwen.ai official	2026-01-22	2026-10-07	hn
A 30B Qwen model walks into a Raspberry Pi and runs in real time	byteshape.com	2026-01-06	2026-10-07	hn
Qwen3-VL	qwen.ai official	2025-09-23	2026-10-07	hn
Qwen3-Omni: Native Omni AI model for text, image and video	github.com	2025-09-22	2026-10-07	hn
Qwen3-Next	qwen.ai official	2025-09-12	2026-10-07	hn
GPT-OSS vs. Qwen3 and a detailed look how things evolved since GPT-2	magazine.sebastianraschka.com	2025-08-10	2026-10-07	hn
Switching to Qwen3-30B-A3B-Instruct-2507 Significantly Improves Local Agent	hexeditreality.com	2025-07-31	2026-10-07	hn
Qwen3-Coder: Agentic coding in the world	qwenlm.github.io official	2025-07-22	2026-10-07	hn
Qwen3: Think deeper, act faster	qwenlm.github.io official	2025-04-28	2026-10-07	hn

Papers

5 papers

Papers 5

arXiv links found in the model card and its tags. "Peer-reviewed" is shown only with evidence: a published venue in OpenAlex (measured) or the authors' own journal reference on arXiv (reported).

Qwen3 Technical Report this model

An Yang, Anfeng Li, Baosong Yang +57 · 2025-05-14 · PDF · HF paper page (349 upvotes)
Qwen2.5-1M Technical Report this model

An Yang, Bowen Yu, Chengyuan Li +25 · 2025-01-26 · PDF · HF paper page (71 upvotes)
RULER: What's the Real Context Size of Your Long-Context Language Models? cited in card

Cheng-Ping Hsieh, Simeng Sun, Samuel Kriman +5 · 2024-04-09 · PDF · HF paper page (42 upvotes)
MInference 1.0: Accelerating Pre-filling for Long-Context LLMs via Dynamic Sparse Attention cited in card

Huiqiang Jiang, Yucheng Li, Chengruidong Zhang +9 · 2024-07-02 · PDF · HF paper page (25 upvotes)
Training-Free Long-Context Scaling of Large Language Models cited in card

Chenxin An, Fei Huang, Jun Zhang +4 · 2024-02-27 · PDF · HF paper page (24 upvotes)

Similar models

3 models

Similar models

GLM-4.7-Flash · 31.2B
same size band (L), also MoE, same task (text-generation), more providers, permissive license
North-Mini-Code-1.0 · 30.5B
same size band (L), also MoE, same task (text-generation), fewer providers, permissive license
Tongyi-DeepResearch-30B-A3B · 30.5B
same size band (L), also MoE, same task (text-generation), permissive license

Sources

4 sources

Sources

Hub snapshot: 2026-10-25 · HF API: 2026-10-24 21:34 UTC · Providers: 2026-10-24 21:44 UTC · Articles: 2026-10-24 21:45 UTC · Summary: 2026-10-07 15:01 UTC · Last verified 2026-10-25

Follow Qwen3-30B-A3B-Instruct-2507

Watch this model

/feeds/models/Qwen--Qwen3-30B-A3B-Instruct-2507.xml(RSS 2.0) · every change to Qwen3-30B-A3B-Instruct-2507
/feeds/families/Qwen--Qwen3.xml(RSS 2.0) · every change across the Qwen3 family

Email alerts are not open yet; the feeds above are the same data. See feeds for the item schema.