Qwen3-0.6B

752M parameters, text in, text out, 40K tokens of context. 1 provider serves it.

Verified 2026-10-25 Rank #30 in LLMs · score 113.5 method v1 Sources 3 Downloads 30d 24.9M · #4 of 619 tracked

Changed this week: 22 new quantized builds on the Hub (503 in all).2026-10-24 · 21 new quantized builds on the Hub (502 in all).2026-10-24

Can I run it?0.5GB at INT4Estimatefits a 16 GB cardmethodCommunity: Qwen3-0.6B — local specs and VRAM →

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

Does it work?23.93on BFCL#24 of 28 measuredleaderboard2026-10-25

Do people use it?24.9Mdownloads, 30 days#4 of 619 we track · top 1%+25 likes, downloads −17% this week1,848 likesHugging Face2026-10-24

Worth reading

Picked by rule from 25 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 · 4 months ago

Good results fine tuning a local LLM like Qwen 3:0.6B to categorize questions

teachmecoolstuff.com 2026-06-21 215 HN pts · 49 comments about the Qwen3 family family

25 best of 25 tracked

The 4 picks above, plus these 21. 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
ArticleQwen3.5 122B and 35B models offer Sonnet 4.5 performance on local computersventurebeat.com · 2026-02-28 · 461 HN pts
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
CodeC inference for Qwen3-ASR 0.6B and 1.7B transcriptions modelsgithub.com · 2026-02-28
CodeShow HN: I wrote inference for Qwen3 0.6B in C/CUDAgithub.com · 2025-09-23 · 4 HN pts
DiscussionQwen 3 0.6B Quantized running at 90tok/s on A19 Pro chiptwitter.com · 2025-10-03 · 7 HN pts
CodeShow HN: Qwen3-TTS integration for Manim animationsgithub.com · 2026-01-24
CodeShow HN: Pure CUDA C Inference for Qwen3 0.6B in One File, No Dependenciesgithub.com · 2025-07-28 · 1 HN pts
ArticleQwen3 Released: How Does It Stack Up?runpod.io · undated
ArticleQwen3-0.6B: Lightweight Open-Source LLM with Thinking Modedev.co · undated
ArticleQwen 3 (Apr 2025) - LLMBasespawn08.github.io · undated
ArticleQwen3-0.6B-Base · Local AI Registrylocal-ai-registry.vercel.app · undated
ArticleQwen3-0.6B — local specs and VRAMalllocalaimodels.com · undated
ArticleShow HN: Duperr – On-device voice cloning using Qwen3-TTS for iOStestflight.apple.com · 2026-01-27
ArticleQwen3 0.6B Free Chat Online - Skywork aiskywork.ai · 2025-11-09

Show all 21

Where to run it

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

#	Cheapest live row	In /1M	Out /1M	Context	Quant	Seen
1	Featherless	—	—	—	—	Oct 25

All 1 provider →

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 1 API provider; heavily run locally (503 quantized builds); a big base for builders (2,085 fine-tunes, adapters and merges).

As a service1Featherless
On own hardware503quantized builds
Built on2,0851,383 fine-tunes · 688 adapters · 14 merges
In apps100+Hugging Face Spaces

Top Spaces running it: k2-fsa/OmniVoice ♥ 1,431 · Qwen/Qwen3-Demo ♥ 868 · pliny-the-prompter/obliteratus ♥ 621 · HuggingFaceFW/finephrase ♥ 285

Will it fit?

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

Fits a 16 GB card at INT4

About 0.5 GB at INT4, 0.9 GB at FP8, 1.8 GB at BF16: weights only, so leave room for the KV cache.

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.

#24 of 28 on BFCL so far

23.93. 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.

8 dated articles and 24 repo commits so far

Teal is the weights (2025-04-27), amber triangles are 8 dated articles (each links to it), grey ticks are repo commits.

What people built on it

2,588 derivatives on the Hub. Which build to run, by runtime: the most downloaded in each class.

2,588 derivatives on the Hub, most of them fine-tunes

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

GGUF MaziyarPanahi/Qwen3-0.6B-GGUF for llama.cpp, Ollama, LM Studio 3.5M downloads MLX mlx-community/Qwen3-0.6B-8bit for Apple silicon 436.6k downloads FP8 Qwen/Qwen3-0.6B-FP8 for vLLM, SGLang 13.5M downloads

All variants →

Reference

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

Benchmarks

2 rows

Benchmarks 2 rows

measured third-party leaderboard row (2)author-reported number from the model card or eval filesNo 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	23.93 percent (overall accuracy)	FC 5 sub-scores Non-Live AST Acc: 71.79 Live Acc: 56.62 Multi Turn Acc: 3.62 Web Search Acc: 1 Memory Acc: 8.6	leaderboard	2026-10-25
measured BFCL · Call tools and APIs	22.38 percent (overall accuracy)	Prompt 5 sub-scores Non-Live AST Acc: 70 Live Acc: 49.37 Multi Turn Acc: 1.38 Web Search Acc: 0.5 Memory Acc: 8.39	leaderboard	2026-10-25

What the card says

1 cited sentence

Summary generated, cited

The model architecture features 28 layers and a context length of 32,768.^[1]^[2]

“Number of Layers: 28”
“Context Length: 32,768”

Generated 2026-10-07 14:59 UTC by parasail-gemma-4-26b-a4b-it; each numeral points at the fact it rests on. Not a verdict.

Use-case evidence

5 use cases

What it is good at 5 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

21 fields

Listed in LLMs

Facts

Qwen3-0.6B has 752M parameters. (safetensors metadata, observed 2026-10-24)
Context length is 40,960 tokens (40K). (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 2025-04-27, last modified 2025-07-26. (HF API, observed 2026-10-24)

Pipeline tag	text-generation
Library	transformers
Architectures	Qwen3ForCausalLM
Model type	qwen3
Dtype	bfloat16
Params by dtype	BF16: 751,632,384
Size band	S
MoE	no
Context length	40,960 (config.json max_position_embeddings)
Papers	arXiv:2505.09388
License	Apache 2.0 (apache-2.0)
Hub snapshot	2026-10-25

Servability

vLLM · SGLang

Servability

vLLM: supported · Qwen3ForCausalLM · 7050ef881ff0835f6890a47db1859a24c65481d6SGLang: supported · Qwen3ForCausalLM · 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 Qwen3-0.6B: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.

BF16 (2 bytes/param)	1.8 GB · Edge (<4 GB)
FP8 (1 byte/param)	0.9 GB · Edge (<4 GB)
INT4 (0.5 bytes/param)	0.5 GB · Edge (<4 GB)

Weights only (params × bytes/param × 1.2 overhead); KV cache and activations extra.

All variants

40 variants

Lineage

Fine-tuned from Qwen/Qwen3-0.6B-Base (same organization) Family: Qwen/Qwen3 2,588 derivatives on the Hub

Base model: Qwen/Qwen3-0.6B-Base

quantized: 503finetune: 1,383adapter: 688merge: 14

Top variants (40 shown, ranked by downloads)

Variant	Relation	Quant	Params	Band	License	Author	Likes	Downloads
Qwen/Qwen3-0.6B-FP8	quantized	-FP8	752M	S	apache-2.0	Qwen	70	13.5M
k2-fsa/OmniVoice	finetune	—	613M	S	—	k2-fsa	1.6k	10.9M
jinaai/jina-reranker-v3	finetune	—	597M	S	cc-by-nc-4.0	jinaai	155	7.2M
MaziyarPanahi/Qwen3-0.6B-GGUF	quantized	GGUF	—	—	—	MaziyarPanahi	15	3.5M
unsloth/Qwen3-0.6B	finetune	—	596M	S	—	unsloth	24	2.9M
unsloth/Qwen3-0.6B-unsloth-bnb-4bit	quantized	-bnb	609M	S	apache-2.0	unsloth	27	1.3M
unsloth/Qwen3-0.6B-GGUF	quantized	GGUF	—	—	apache-2.0	unsloth	142	1.3M
farbodtavakkoli/OTel-Reranker-0.6B	finetune	—	596M	S	apache-2.0	farbodtavakkoli	2	1.2M
farbodtavakkoli/OTel-LLM-0.6B-IT	finetune	—	—	—	apache-2.0	farbodtavakkoli	0	1.2M
Qwen/Qwen3-0.6B-GGUF	quantized	GGUF	—	—	apache-2.0	Qwen	103	985.3k
Skywork/Skywork-Reward-V2-Qwen3-0.6B	finetune	—	596M	S	apache-2.0	Skywork	14	831.5k
ggml-org/Qwen3-0.6B-GGUF	quantized	GGUF	—	—	apache-2.0	ggml-org	21	568.6k
mlx-community/Qwen3-0.6B-8bit	quantized	-8bit	596M	S	apache-2.0	mlx-community	9	436.6k
mlx-community/Qwen3-0.6B-4bit	quantized	-4bit	596M	S	apache-2.0	mlx-community	18	425.4k
bartowski/Qwen_Qwen3-0.6B-GGUF	quantized	GGUF	—	—	—	bartowski	28	389k
Qwen/Qwen3Guard-Stream-0.6B	finetune	—	597M	S	apache-2.0	Qwen	51	257.9k
litert-community/Qwen3-0.6B	quantized	—	—	—	apache-2.0	litert-community	26	255.5k
rogue-security/prompt-injection-jailbreak-sentinel-v2	finetune	—	596M	S	other	rogue-security	58	216.9k
bd2lcco/Qwen3-0.6B-finetuned	finetune	—	596M	S	apache-2.0	bd2lcco	1	202.6k
unsloth/Qwen3-0.6B-bnb-4bit	quantized	-bnb	610M	S	—	unsloth	4	173.9k
jinaai/jina-reranker-v3.5	finetune	—	597M	S	cc-by-nc-4.0	jinaai	56	159.4k
charent/self_cognition_Alice	adapter	—	—	—	—	charent	1	111.4k
onnx-community/Qwen3-0.6B-ONNX	quantized	—	—	—	—	onnx-community	44	102k
lmstudio-community/Qwen3-0.6B-GGUF	quantized	GGUF	—	—	apache-2.0	lmstudio-community	13	88.7k
geoffmunn/Qwen3-0.6B-f16	quantized	GGUF	—	—	apache-2.0	geoffmunn	1	67k
lmstudio-community/Qwen3-0.6B-MLX-bf16	finetune	MLX	596M	S	apache-2.0	lmstudio-community	2	59.5k
lmstudio-community/Qwen3-0.6B-MLX-4bit	quantized	MLX	596M	S	apache-2.0	lmstudio-community	0	59.1k
RedHatAI/Qwen3-0.6B-quantized.w4a16	quantized	-quant	765M	S	apache-2.0	RedHatAI	0	54.7k
Qwen/Qwen3-0.6B-GPTQ-Int8	quantized	GPTQ	596M	S	apache-2.0	Qwen	17	53.7k
mlc-ai/Qwen3-0.6B-q4f16_1-MLC	quantized	—	—	—	—	mlc-ai	1	53.9k
Orion-zhen/Qwen3-0.6B-AWQ	quantized	AWQ	596M	S	gpl-3.0	Orion-zhen	6	42.9k
dllm-hub/Qwen3-0.6B-diffusion-mdlm-v0.1	finetune	—	752M	S	apache-2.0	dllm-hub	16	37.1k
divelab/OPDLM-0.6B	finetune	—	596M	S	mit	divelab	0	33k
superwhisper/s1-mini	finetune	—	752M	S	other	superwhisper	379	12.3k
codelion/Qwen3-0.6B-accuracy-recovery-lora	adapter	—	—	—	apache-2.0	codelion	5	28.4k
RedHatAI/Qwen3-0.6B-FP8-dynamic	quantized	-FP8	752M	S	apache-2.0	RedHatAI	1	27.7k
OpenVINO/Qwen3-0.6B-int4-ov	quantized	int4	—	—	apache-2.0	OpenVINO	3	26.3k
doodod/Turn-Detector-Qwen3-0.6B	quantized	—	596M	S	mit	doodod	0	25.2k
dllm-hub/Qwen3-0.6B-diffusion-bd3lm-v0.1	finetune	—	752M	S	apache-2.0	dllm-hub	12	23.8k
reaperdoesntknow/Qemma-sft	finetune	—	—	—	osl-3.0	reaperdoesntknow	0	24.3k

Repository changelog

24 commits

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

Repository changelog 24 commits

Date	Commit	SHA
2025-07-26	Create LICENSE	c1899de
2025-05-21	Update README.md	e6de914
2025-05-19	update tokenizer_config.json	7e4ae26
2025-04-29	Update README.md	6130ef3
2025-04-28	Update README.md	a9c98e6
2025-04-28	Update README.md	f4863e3
2025-04-28	Update README.md	7523470
2025-04-28	Update README.md	715e27a
2025-04-28	Update tokenizer_config.json	16706fc
2025-04-28	Delete special_tokens_map.json	a819064
2025-04-28	Update tokenizer_config.json	e178a19
2025-04-28	Delete added_tokens.json	271af0b
2025-04-28	Update README.md	cfa73ab
2025-04-28	Update README.md	ea7997d
2025-04-28	Update README.md	5cdf60d
2025-04-28	Update README.md	6fad4ed
2025-04-28	Update README.md	ec0dcdf
2025-04-28	Update README.md	a08cec3
2025-04-28	Delete configuration.json	1c2706d
2025-04-28	Delete .mv	749a61e
2025-04-28	Delete .msc	9adc26d
2025-04-28	Update README.md	66b95ce
2025-04-28	Upload folder using huggingface_hub	167b810
2025-04-27	initial commit	e15642e

All providers

1 provider · 1 endpoint

Where to run 1 providers · 1 endpoints

Hosted endpoints observed for Qwen3-0.6B, sorted by input price (USD per million tokens). Each row names its observer and date; prices change daily, so check the source before quoting.

Provider	Model ref	Context	Quant	$/M in	$/M out	tok/s	TTFT	Tools	Status	Observed
Featherless hf	Qwen/Qwen3-0.6B	—	—	—	—	—	—	—	live	2026-10-25

Articles and coverage

25 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.

Analysis8

Good results fine tuning a local LLM like Qwen 3:0.6B to categorize questions

teachmecoolstuff.com2026-06-21HN front pageabout the Qwen3 familyHN 215 pts · 49 comments

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.

Show HN: I wrote inference for Qwen3 0.6B in C/CUDA

github.com2025-09-23authorityHN 4 pts · 0 comments

Over 1k tok/s on an RTX 5090 with Qwen3 0.6B

blog.alpindale.net2026-02-09HN 2 pts · 1 comments

Show HN: Pure CUDA C Inference for Qwen3 0.6B in One File, No Dependencies

github.com2025-07-28authorityHN 1 pts · 0 comments

Qwen3-0.6B by Qwen - Overview, Benchmarks & Hardware | Q4KM.ai

q4km.ai

Qwen3 Released: How Does It Stack Up?

runpod.ioabout the Qwen3 family

Community1

Qwen 3 0.6B Quantized running at 90tok/s on A19 Pro chip

twitter.com2025-10-03HN 7 pts · 1 comments

All coverage tracked for this model (25) — append-only, nothing is ever removed

Title	Source	Published	First seen	Signals
Qwen3-0.6B by Qwen - Overview, Benchmarks & Hardware \| Q4KM.ai	q4km.ai	—	2026-10-07	exa
Qwen3 Released: How Does It Stack Up?	runpod.io	—	2026-10-07	exa
Qwen3-0.6B: Lightweight Open-Source LLM with Thinking Mode	dev.co	—	2026-10-07	exa
Qwen 3 (Apr 2025) - LLMBase	spawn08.github.io	—	2026-10-07	exa
Qwen3-0.6B-Base · Local AI Registry	local-ai-registry.vercel.app	—	2026-10-07	exa
Qwen3-0.6B — local specs and VRAM	alllocalaimodels.com	—	2026-10-07	exa
Good results fine tuning a local LLM like Qwen 3:0.6B to categorize questions	teachmecoolstuff.com	2026-06-21	2026-10-07	hn
Qwen3.5 122B and 35B models offer Sonnet 4.5 performance on local computers	venturebeat.com	2026-02-28	2026-10-07	hn
C inference for Qwen3-ASR 0.6B and 1.7B transcriptions models	github.com	2026-02-28	2026-10-07	hn
Over 1k tok/s on an RTX 5090 with Qwen3 0.6B	blog.alpindale.net	2026-02-09	2026-10-07	hn
Qwen3-Coder-Next	qwen.ai official	2026-02-03	2026-10-07	hn
Show HN: Duperr – On-device voice cloning using Qwen3-TTS for iOS	testflight.apple.com	2026-01-27	2026-10-07	hn
Qwen3-Max-Thinking	qwen.ai official	2026-01-26	2026-10-07	hn
Show HN: Qwen3-TTS integration for Manim animations	github.com	2026-01-24	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
Qwen3 0.6B Free Chat Online - Skywork ai	skywork.ai	2025-11-09	2026-10-07	exa
Qwen 3 0.6B Quantized running at 90tok/s on A19 Pro chip	twitter.com	2025-10-03	2026-10-07	hn
Qwen3-VL	qwen.ai official	2025-09-23	2026-10-07	hn
Show HN: I wrote inference for Qwen3 0.6B in C/CUDA	github.com	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
Show HN: Pure CUDA C Inference for Qwen3 0.6B in One File, No Dependencies	github.com	2025-07-28	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

1 paper

Papers 1

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)

Similar models

3 models

Similar models

Qwen3Guard-Gen-0.6B · 752M
same size band (S), same task (text-generation), permissive license
gpt2-large · 812M
same size band (S), same task (text-generation), permissive license
bitnet-b1.58-2B-4T · 850M
same size band (S), same task (text-generation), permissive license

Sources

3 sources

Sources

Hub snapshot: 2026-10-25 · HF API: 2026-10-24 21:32 UTC · Providers: 2026-10-24 21:44 UTC · Articles: 2026-10-24 21:45 UTC · Summary: 2026-10-07 14:59 UTC · Last verified 2026-10-25

Follow Qwen3-0.6B

Watch this model

/feeds/models/Qwen--Qwen3-0.6B.xml(RSS 2.0) · every change to Qwen3-0.6B
/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.