Model Follower.

Up to date

Rebuilt
2026-10-25 06:54 UTC
Release radar
117 on watch
Hub snapshot
2026-10-25
Re-checked
daily · every number sourced and dated

Models › Qwen1.5 family

Qwen1.5-0.5B

Hugging Face ↗ Follow changes

Qwen Research / non-commercial (tongyi-qianwen-research) out 3 years ago

620M parameters, text in, text out, 32K tokens of context. 1 provider serves it.

Verified 2026-10-25 Rank #311 in LLMs · score 83 method v1 Sources 3 Downloads 30d 36.5k · #407 of 619 tracked

Can I run it?0.4GB at INT4Estimatefits a 16 GB cardmethod
Can I use it?Research / non-commercial (tongyi-qianwen-research)Non-commercial onlylicence file2026-10-24
Do people use it?36.5kdownloads, 30 days#407 of 619 we track · below the median+2 likes, downloads −4% this week182 likesHugging Face2026-10-24

Worth reading

Picked by rule from 10 tracked pieces: the lab's own post, the most shared, a field report, the latest.

Latest · 3 years ago

Qwen1.5-Moe: Matching 7B Model Performance with 1/3 Activated Parameters

Qwen1.5-MoE-A2.7B matches 7B model performance with only 2.7 billion activated parameters using a 64-expert configuration.

  • The model partitions a single FFN into segments to generate 64 experts without increasing the total parameter count.
  • Training from scratch is inefficient, so the team repurposes the existing Qwen-1.8B model to initialize Qwen1.5-MoE-A2.7B.
qwenlm.github.io 2024-03-28 104 HN pts · 10 comments 1.5k words about the Qwen1.5 family family

10 best of 10 tracked

The 2 picks above, plus these 8. Top is the radar's score: discussion, source authority and the lab's own posts, fading with age.

  1. ArticleQwen1.5-110Bqwenlm.github.io · 2024-04-26 · official
  2. ArticleQwen1.5-32B: Fitting the Capstone of the Qwen1.5 Language Model Seriesqwenlm.github.io · 2024-04-05 · official
  3. ArticleQwen1.5qwenlm.github.io · 2024-02-05 · 5 HN pts · official
  4. ArticleTiny LLM Reviews: Qwen 1.5 0.5bkyle.works · undated
  5. DiscussionQwen1.5-110B Previewtwitter.com · 2024-04-20
  6. ArticleRun Alicloud's Qwen LLM with a 2MB cross GPU/CPU rust appsecondstate.io · 2024-02-28
  7. Article通义千问Qwen1.5系列大模型(低参数版本)实验_AIGC_原野AI-魔乐社区modelers.csdn.net · 2024-06-07
  8. ArticleIntroducing Qwen1.5 | Qiyuan.Techqiyuan-tech.github.io · 2024-02-04

Where to run it

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

#Cheapest live rowIn /1MOut /1MContextQuantSeen
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; 27 quantized builds for local use; a big base for builders (32,507 fine-tunes, adapters and merges).

  • As a service1Featherless
  • On own hardware27quantized builds
  • Built on32,50732 fine-tunes · 32,473 adapters · 2 merges
  • In apps39Hugging Face Spaces

Top Spaces running it: fromziro/SLM_Regression_Line ♥ 16 · Justinrune/LLaMA-Factory ♥ 12 · aelitta/BioMistral_gradio ♥ 2 · rockyripple04/Qwen-Qwen1.5-0.5B ♥ 1

Will it fit?

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

Fits a 16 GB card at INT4

About 0.4 GB at INT4, 0.7 GB at FP8, 1.5 GB at BF16: weights only, so leave room for the KV cache.

16 GB card64 GB Mac2×80 GB8×80 GBINT4: about 0.4 GB of weights (estimate; KV cache and activations extra)INT40.4 GBFP8: about 0.7 GB of weights (estimate; KV cache and activations extra)FP80.7 GBBF16: about 1.5 GB of weights (estimate; KV cache and activations extra)BF161.5 GB
16 GB card64 GB Mac2×80 GB8×80 GBINT4: about 0.4 GB of weights (estimate; KV cache and activations extra)INT40.4 GBFP8: about 0.7 GB of weights (estimate; KV cache and activations extra)FP80.7 GBBF16: about 1.5 GB of weights (estimate; KV cache and activations extra)BF161.5 GB

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

The story so far

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

5 dated articles and 23 repo commits so far

Teal is the weights (2024-01-22), amber triangles are 5 dated articles (each links to it), grey ticks are repo commits.

FebMarAprMayJunJulAugSepOctNovDecJanFebMarAprMayJunJulAugSepOctNovDecJanFebMarAprMayJunJulAugSepOct2024-04-05: 1 repo commit2024-04-04: 1 repo commit2024-02-05: 3 repo commits2024-02-03: 4 repo commits2024-01-31: 3 repo commits2024-01-30: 2 repo commits2024-01-29: 6 repo commits2024-01-27: 1 repo commit2024-01-22: 2 repo commitsQwen1.5-Moe: Matching 7B Model Performance with 1/3 Activated Parameters · 2024-03-28Qwen1.5 · 2024-02-05Complete tutorial: Run Qwen LLM across your local machines without Python/C++ · 2024-02-22Introducing Qwen1.5 | Qiyuan.Tech · 2024-02-04通义千问Qwen1.5系列大模型(低参数版本)实验_AIGC_原野AI-魔乐社区 · 2024-06-07Weights 2024-01-22 · 2024-01-22Weights 2024-01-22now · 1 provider
FebMarAprMayJunJulAugSepOctNovDecJanFebMarAprMayJunJulAugSepOctNovDecJanFebMarAprMayJunJulAugSepOct2024-04-05: 1 repo commit2024-04-04: 1 repo commit2024-02-05: 3 repo commits2024-02-03: 4 repo commits2024-01-31: 3 repo commits2024-01-30: 2 repo commits2024-01-29: 6 repo commits2024-01-27: 1 repo commit2024-01-22: 2 repo commitsQwen1.5-Moe: Matching 7B Model Performance with 1/3 Activated Parameters · 2024-03-28Qwen1.5 · 2024-02-05Complete tutorial: Run Qwen LLM across your local machines without Python/C++ · 2024-02-22Introducing Qwen1.5 | Qiyuan.Tech · 2024-02-04通义千问Qwen1.5系列大模型(低参数版本)实验_AIGC_原野AI-魔乐社区 · 2024-06-07Weights 2024-01-22 · 2024-01-22Weights 2024-01-22now · 1 provider

What people built on it

32,534 derivatives on the Hub. Which build to run, by runtime: the most downloaded in each class.

32,534 derivatives on the Hub, most of them adapters

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

27 quantized (Hugging Face model tree)32 fine-tunes (Hugging Face model tree)2 merges (Hugging Face model tree)32,473 adapters (Hugging Face model tree)27 quantized32 fine-tunes2 merges32,473 adapters
27 quantized (Hugging Face model tree)32 fine-tunes (Hugging Face model tree)2 merges (Hugging Face model tree)32,473 adapters (Hugging Face model tree)27 quantized32 fine-tunes2 merges32,473 adapters
GGUF mradermacher/Qwen1.5-0.5B-i1-GGUF for llama.cpp, Ollama, LM Studio 5.6k downloads

All variants →

Reference

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

What the card says

1 cited sentence

Summary generated, cited

Qwen1.5-0.5B is a 620M parameter decoder-only language model designed for text-generation tasks.[1][2]

  1. parameters human
  2. “decoder-only language model”

Generated 2026-10-07 15:08 UTC by parasail-gemma-4-26b-a4b-it; each numeral points at the fact it rests on. Not a verdict.

Use-case evidence

2 use cases

What it is good at 2 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

22 fields

Listed in LLMs

Facts

Pipeline tagtext-generation
Librarytransformers
ArchitecturesQwen2ForCausalLM
Model typeqwen2
Dtypebfloat16
Params by dtypeBF16: 619,570,176
Size bandS
MoEno
Context length32,768 (config.json max_position_embeddings)
Languagesen
PapersarXiv:2309.16609
LicenseResearch / non-commercial (tongyi-qianwen-research) (other)
Hub snapshot2026-10-25

Servability

vLLM · SGLang

Servability

vLLM: supported · Qwen2ForCausalLM · 7050ef881ff0835f6890a47db1859a24c65481d6SGLang: supported · Qwen2ForCausalLM · 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 Qwen1.5-0.5B: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.

BF16 (2 bytes/param)1.5 GB · Edge (<4 GB)
FP8 (1 byte/param)0.7 GB · Edge (<4 GB)
INT4 (0.5 bytes/param)0.4 GB · Edge (<4 GB)

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

All variants

40 variants

Lineage

Original model Family: Qwen/Qwen1.5 32,534 derivatives on the Hub

finetune: 32adapter: 32,473quantized: 27merge: 2

Top variants (40 shown, ranked by downloads)

Relation
40 rows
Variant Relation Quant Params Band License Author Likes Downloads
numind/NuExtract-tiny finetune — 464M S mit numind 45 486.8k
ahsbdcpu/task-17-Qwen-Qwen1.5-0.5B adapter — — — — ahsbdcpu 0 33.4k
sail/Sailor-0.5B finetune — 620M S apache-2.0 sail 10 27.8k
Hugging-Face-User-2024/task-15-Qwen-Qwen1.5-0.5B adapter — — — — Hugging-Face-User-2024 0 13.1k
jialei12138/task-13-Qwen-Qwen1.5-0.5B adapter — — — — jialei12138 0 7.5k
xueyj/task-13-Qwen-Qwen1.5-0.5B adapter — — — — xueyj 0 6.9k
lucifer-ms/task-3-Qwen-Qwen1.5-0.5B adapter — — — — lucifer-ms 0 6k
xiny47661/task-15-Qwen-Qwen1.5-0.5B adapter — — — — xiny47661 0 5.7k
mradermacher/Qwen1.5-0.5B-i1-GGUF quantized GGUF — — other mradermacher 0 5.6k
kromer9margarete337/task-13-Qwen-Qwen1.5-0.5B adapter — — — — kromer9margarete337 0 5.4k
xueyj/task-15-Qwen-Qwen1.5-0.5B adapter — — — — xueyj 0 5.1k
l8623001/task-1-Qwen-Qwen1.5-0.5B adapter — — — — l8623001 0 5k
xueyj/task-17-Qwen-Qwen1.5-0.5B adapter — — — — xueyj 0 5k
trl-lib/qwen1.5-0.5b-sft finetune — 464M S other trl-lib 0 4.7k
brittlewis12/Qwen1.5-0.5B-OpenHermes-2.5-GGUF quantized GGUF — — other brittlewis12 2 4.2k
888facaiboy/task-13-Qwen-Qwen1.5-0.5B adapter — — — — 888facaiboy 0 4k
dogssss/task-13-Qwen-Qwen1.5-0.5B adapter — — — — dogssss 0 3.9k
Hugging-Face-User-2024/task-5-Qwen-Qwen1.5-0.5B adapter — — — — Hugging-Face-User-2024 0 3.8k
mradermacher/Qwen1.5-0.5B-GGUF quantized GGUF — — other mradermacher 0 3.7k
WhereIsAI/ese-qwen-0.5b-nli adapter — — — other WhereIsAI 0 2.9k
tensorblock/Qwen1.5-0.5B-GGUF quantized GGUF — — other tensorblock 0 2.6k
geoplus/task-5-Qwen-Qwen1.5-0.5B adapter — — — — geoplus 0 2.1k
wuyanzu4692/task-8-Qwen-Qwen1.5-0.5B adapter — — — — wuyanzu4692 0 1.9k
DevQuasar-8/Qwen.Qwen1.5-0.5B-GGUF quantized GGUF — — — DevQuasar-8 0 1.7k
ecamli/task-2-Qwen-Qwen1.5-0.5B adapter — — — — ecamli 0 1.6k
Xenova/Qwen1.5-0.5B quantized — — — — Xenova 1 1.2k
lucifer-ms/task-1-Qwen-Qwen1.5-0.5B adapter — — — — lucifer-ms 0 890
WildSage/task-3-Qwen-Qwen1.5-0.5B adapter — — — — WildSage 0 809
brandolorian/answer-Qwen-stioning finetune — 464M S other brandolorian 0 784
PursuitOfDataScience/Argonne-Qwen1.5-0.5B-think finetune — 464M S apache-2.0 PursuitOfDataScience 0 712
jabari1993/task-1-Qwen-Qwen1.5-0.5B adapter — — — — jabari1993 0 580
feelsogoodeasy/Qwen-Qwen1.5-0.5B-1727329976 adapter — — — — feelsogoodeasy 0 560
dioufam02/task-3-Qwen-Qwen1.5-0.5B adapter — 464M S — dioufam02 0 491
eoinedge/edgeai-docs-qwen2.5-coder-0.5b-lora adapter GGUF — — apache-2.0 eoinedge 0 477
vilm/Quyen-SE-4e finetune — 620M S other vilm 0 417
gerbyk/task-15-Qwen-Qwen1.5-0.5B adapter — — — — gerbyk 0 414
zhongwen8868/task-13-Qwen-Qwen1.5-0.5B adapter — — — — zhongwen8868 0 391
itlwas/Qwen1.5-0.5B-Q4_K_M-GGUF quantized GGUF — — other itlwas 0 360
kulia-moon/Lily-Qwen1.5-0.5B finetune — 464M S mit kulia-moon 0 327
anseeyeon/task-7-Qwen-Qwen1.5-0.5B adapter — — — — anseeyeon 0 316

Repository changelog

23 commits

Repository changelog 23 commits

DateCommitSHA
2024-04-05 Update README.md 8f445e3
2024-04-04 update tokenizer_config.json 8be5956
2024-02-05 update readme fedce23
2024-02-05 update readme b2108e9
2024-02-05 Update README.md f7d8e35
2024-02-03 Update config.json 4196469
2024-02-03 Update config.json 283c238
2024-02-03 Update generation_config.json 700ded3
2024-02-03 Update generation_config.json d0dd2d2
2024-01-31 Update README.md 2649bd4
2024-01-31 Update README.md 6f9a97e
2024-01-31 Create LICENSE 42c72d8
2024-01-30 Update generation_config.json bdb9b42
2024-01-30 Update config.json a9a7d3a
2024-01-29 Update README.md 37e47e7
2024-01-29 Update README.md 82cec1d
2024-01-29 Update README.md 3f1e072
2024-01-29 Delete special_tokens_map.json 8518547
2024-01-29 Delete added_tokens.json b8aaae8
2024-01-29 Upload tokenizer_config.json 8900548
2024-01-27 Update README.md 44fa223
2024-01-22 Upload folder using huggingface_hub 61dc162
2024-01-22 initial commit 5f76bf3

All providers

1 provider · 1 endpoint

Where to run 1 providers · 1 endpoints

Hosted endpoints observed for Qwen1.5-0.5B, sorted by input price (USD per million tokens). Each row names its observer and date; prices change daily, so check the source before quoting.

ProviderModel refContextQuant$/M in$/M outtok/sTTFTToolsStatusObserved
Featherless
hf
 Qwen/Qwen1.5-0.5B — — — — — — — live 2026-10-25

Articles and coverage

10 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.

Official2

Qwen1.5-Moe: Matching 7B Model Performance with 1/3 Activated Parameters
qwenlm.github.io2024-03-28officialauthorityHN front pageabout the Qwen1.5 familyHN 104 pts · 10 comments1,495 words
Qwen1.5-MoE-A2.7B matches 7B model performance with only 2.7 billion activated parameters using a 64-expert configuration. The model incorporates 4 shared experts alongside 60 routing experts and requires source installation via git for Hugging Face support. It achieves a 75% decrease in training expenses and accelerates inference speed by a factor of 1.74 compared to Qwen1.5-7B.written, reviewed, cited
  • The model partitions a single FFN into segments to generate 64 experts without increasing the total parameter count.
  • Training from scratch is inefficient, so the team repurposes the existing Qwen-1.8B model to initialize Qwen1.5-MoE-A2.7B.
  • The Qwen1.5-MoE-A2.7B-Chat-GPTQ-Int4 quantized variant is available, but AWQ support is currently not supported.
Qwen1.5
qwenlm.github.io2024-02-05officialauthorityabout the Qwen1.5 familyHN 5 pts · 2 comments3,793 words
Qwen1.5 is a new open-source model series released before the Chinese New Year, offering base and chat models in six sizes ranging from 0.5B to 110B parameters. The Qwen1.5-72B model outperforms Llama2-70B across all benchmarks, and the series features improved multilingual and human-alignment capabilities. All models now uniformly support a context length of up to 32768 tokens, and the Qwen1.5-72B-Chat model is recommended for initial testing.written, reviewed, cited

Analysis4

Tiny LLM Reviews: Qwen 1.5 0.5b
kyle.works1,497 words
Qwen 1.5 0.5b is a 500 million parameter model effective for document summarization and simple logic but struggles with programming and factual accuracy. It generates a business email in 5 minutes and code in 10+ minutes, while incorrectly stating New Delhi's population as 1.3 billion instead of 28 million. The model also misreports Seattle's annual rainfall as 998mm versus New Delhi's 774mm.written, reviewed, cited
  • The model fails to reverse a linked list because it returns a Node object missing an accumulator function.
  • Generation time increases with chat length because the system sends the whole text of the chat to the model each time.
  • The model confuses India with New Delhi twice, once where language was purposely vague and once where language is clear.
Complete tutorial: Run Qwen LLM across your local machines without Python/C++
secondstate.io2024-02-22HN 3 pts · 0 comments629 words
Introducing Qwen1.5 | Qiyuan.Tech
qiyuan-tech.github.io2024-02-04about the Qwen1.5 family3,667 words
通义千问Qwen1.5系列大模型(低参数版本)实验_AIGC_原野AI-魔乐社区
modelers.csdn.net2024-06-07about the Qwen1.5 family176 words
All coverage tracked for this model (10) — append-only, nothing is ever removed
TitleSourcePublishedFirst seenSignals
Tiny LLM Reviews: Qwen 1.5 0.5bkyle.works—2026-10-23exa
通义千问Qwen1.5系列大模型(低参数版本)实验_AIGC_原野AI-魔乐社区modelers.csdn.net2024-06-072026-10-23exa
Qwen1.5-110Bqwenlm.github.io official2024-04-262026-10-07hn
Qwen1.5-110B Previewtwitter.com2024-04-202026-10-07hn
Qwen1.5-32B: Fitting the Capstone of the Qwen1.5 Language Model Seriesqwenlm.github.io official2024-04-052026-10-07hn
Qwen1.5-Moe: Matching 7B Model Performance with 1/3 Activated Parametersqwenlm.github.io official2024-03-282026-10-07hn
Run Alicloud's Qwen LLM with a 2MB cross GPU/CPU rust appsecondstate.io2024-02-282026-10-07hn
Complete tutorial: Run Qwen LLM across your local machines without Python/C++secondstate.io2024-02-222026-10-07hn
Qwen1.5qwenlm.github.io official2024-02-052026-10-07hn
Introducing Qwen1.5 | Qiyuan.Techqiyuan-tech.github.io2024-02-042026-10-23exa

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).

Similar models

3 models

Similar models

  • Qwen3-0.6B-Base · 596M
    same size band (S), same task (text-generation), permissive license
  • Qwen3-0.6B · 752M
    same size band (S), same task (text-generation), permissive license
  • bloomz-560m · 559M
    same size band (S), same task (text-generation)

Sources

3 sources

Sources

Hub snapshot: 2026-10-25 · HF API: 2026-10-24 21:38 UTC · Providers: 2026-10-24 21:44 UTC · Articles: 2026-10-24 21:46 UTC · Summary: 2026-10-07 15:08 UTC · Last verified 2026-10-25

Follow Qwen1.5-0.5B

Watch this model

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