Qwen2-1.5B

1.54B parameters, text in, text out, 128K tokens of context. 1 provider serves it.

Verified 2026-10-25 Rank #454 in LLMs · score 77.9 method v1 Sources 3 Downloads 30d 124.4k · #293 of 619 tracked

Can I run it?0.9GB at INT4Estimatefits a 16 GB cardmethod

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

Do people use it?124.4kdownloads, 30 days#293 of 619 we track · top 47%+2 likes, downloads −4% this week114 likesHugging Face2026-10-24

Worth reading

Picked by rule from 7 tracked pieces: the lab's own post, the most shared, a field report, the latest.

Most discussed · 29 HN pts

Alibaba Qwen2.5-Omni-7B: Open Source End-to-End Multimodal AI Model

alizila.com 2025-03-27 29 HN pts · 3 comments about the Qwen2 family family

Latest · 54 days ago

Puro-2B: Poor Lab's Qwen2-1.5B Trained on RTX 5090 within $5090

Poor Lab releases Puro-2B, an open pretraining recipe that trains Qwen2-1.5B models from scratch on up to 1.4 trillion tokens with FP8 precision on consumer-grade RTX 5090 GPUs.

arxiv.org 2026-09-01 3 HN pts 485 words

7 best of 7 tracked

The 2 picks above, plus these 5. Top is the radar's score: discussion, source authority and the lab's own posts, fading with age.

ArticleQwen2 LLM Releasedqwenlm.github.io · 2024-06-06 · 261 HN pts · official
ArticleQwen2-Mathqwenlm.github.io · 2024-08-08 · 128 HN pts · official
CodeLlama.cpp Now Supports Qwen2-VL (Vision Language Model)github.com · 2024-12-14
ArticleAn Analysis of Chinese LLM Censorship and Bias with Qwen 2 Instructhuggingface.co · 2024-06-09 · 16 HN pts
ArticleQwen2.5-Max: The Next Leap in AI Innovationmedium.com · 2025-01-29 · 17 HN pts

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; 23 quantized builds for local use; a big base for builders (1,518 fine-tunes, adapters and merges).

As a service1Featherless
On own hardware23quantized builds
Built on1,51888 fine-tunes · 1,429 adapters · 1 merges
In apps18Hugging Face Spaces

Top Spaces running it: filapro/cad-recode ♥ 22 · fromziro/SLM_Regression_Line ♥ 16

Will it fit?

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

Fits a 16 GB card at INT4

About 0.9 GB at INT4, 1.9 GB at FP8, 3.7 GB at BF16: weights only, so leave room for the KV cache.

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

The story so far

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

6 dated articles and 5 repo commits so far

Teal is the weights (2024-05-31), amber triangles are 6 dated articles (each links to it), grey ticks are repo commits.

What people built on it

1,541 derivatives on the Hub. Which build to run, by runtime: the most downloaded in each class.

1,541 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.

GGUF legraphista/Qwen2-1.5B-IMat-GGUF for llama.cpp, Ollama, LM Studio 46.7k downloads MLX TheBlueObserver/Qwen2.5-1.5B-Instruct-4bit-128-GPTQ-MLX-LinearOnly for Apple silicon 585 downloads

All variants →

Reference

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

Benchmarks

14 rows

Benchmarks 14 rows

measured third-party leaderboard rowauthor-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
author-reported ARC Challenge · Chat and follow instructions	43.9 percent	—	model author	2026-10-07
author-reported BBH · Reason through hard problems	37.2 percent	—	model author	2026-10-07
author-reported C-Eval · Work across languages	70.6 percent	—	model author	2026-10-07
author-reported CMMLU · Work across languages	70.3 percent	—	model author	2026-10-07
author-reported GSM8K · Reason through hard problems	58.5 percent	—	model author	2026-10-07
author-reported HellaSwag · Chat and follow instructions	66.6 percent	—	model author	2026-10-07
author-reported HumanEval · Write code with agents	31.1 percent	—	model author	2026-10-07
author-reported MATH-500 · Reason through hard problems	21.7 percent	—	model author	2026-10-07
author-reported MBPP · Write code with agents	37.4 percent	—	model author	2026-10-07
author-reported MMLU · Reason through hard problems	56.5 percent	—	model author	2026-10-07
author-reported MMLU-Pro · Reason through hard problems	21.8 percent	—	model author	2026-10-07
author-reported Theorem QA	15 percent	—	model author	2026-10-07
author-reported TruthfulQA · Chat and follow instructions	45.9 percent	—	model author	2026-10-07
author-reported Winogrande · Chat and follow instructions	66.2 percent	—	model author	2026-10-07

What the card says

2 cited sentences

Summary generated, cited

Qwen2-1.5B is a 1.54B parameter base language model from the Qwen2 series.^[1]^[2] The model uses a Transformer architecture with SwiGLU activation and group query attention.^[3]

parameters human
“This repo contains the 1.5B Qwen2 base language model.”
“It is based on the Transformer architecture with SwiGLU a…”

Generated 2026-10-07 15:10 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

21 fields

Listed in LLMs

Facts

Qwen2-1.5B has 1.54B parameters. (safetensors metadata, observed 2026-10-24)
Context length is 131,072 tokens (128K). (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 2024-05-31, last modified 2024-06-06. (HF API, observed 2026-10-24)

Pipeline tag	text-generation
Library	transformers
Architectures	Qwen2ForCausalLM
Model type	qwen2
Dtype	bfloat16
Params by dtype	BF16: 1,543,714,304
Size band	S
MoE	no
Context length	131,072 (config.json max_position_embeddings)
Languages	en
License	Apache 2.0 (apache-2.0)
Hub snapshot	2026-10-25

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 Qwen2-1.5B: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.

BF16 (2 bytes/param)	3.7 GB · Edge (<4 GB)
FP8 (1 byte/param)	1.9 GB · Edge (<4 GB)
INT4 (0.5 bytes/param)	0.9 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/Qwen2 1,541 derivatives on the Hub

finetune: 88quantized: 23adapter: 1,429merge: 1

Top variants (40 shown, ranked by downloads)

Variant	Relation	Quant	Params	Band	License	Author	Likes	Downloads
filapro/cad-recode-v1.5	finetune	—	1.54B	S	cc-by-nc-4.0	filapro	10	80.2k
legraphista/Qwen2-1.5B-IMat-GGUF	quantized	GGUF	—	—	apache-2.0	legraphista	0	46.7k
bartowski/Replete-Coder-Qwen2-1.5b-GGUF	quantized	GGUF	—	—	apache-2.0	bartowski	14	23.6k
QuantFactory/Qwen2-1.5B-GGUF	quantized	GGUF	—	—	apache-2.0	QuantFactory	3	14.9k
mradermacher/Qwen2-1.5B-GGUF	quantized	GGUF	—	—	apache-2.0	mradermacher	1	14.5k
LenguajeNaturalAI/leniachat-qwen2-1.5B-v0	finetune	—	1.54B	S	apache-2.0	LenguajeNaturalAI	21	8k
QuantFactory/leniachat-qwen2-1.5B-v0-GGUF	quantized	GGUF	—	—	apache-2.0	QuantFactory	4	6.8k
dphn/dolphin-2.9.3-qwen2-1.5b	finetune	—	1.54B	S	apache-2.0	dphn	13	5.9k
filapro/cad-recode	finetune	—	1.54B	S	cc-by-nc-4.0	filapro	20	5.3k
QuantFactory/oxy-1-micro-GGUF	quantized	GGUF	—	—	apache-2.0	QuantFactory	4	4.8k
archit28/Qwen2-1.5B-Q4_K_S-GGUF	quantized	GGUF	—	—	apache-2.0	archit28	0	4.3k
afrideva/Qwen2-1.5B-GGUF	quantized	GGUF	—	—	apache-2.0	afrideva	0	4.3k
artificialguybr/QWEN-2-1.5B-Synthia-I	finetune	—	1.54B	S	apache-2.0	artificialguybr	2	4.1k
artificialguybr/QWEN-2-1.5B-Synthia-II-Redmond	finetune	—	—	—	apache-2.0	artificialguybr	2	4.1k
Deeps03/qwen2-1.5b-log-classifier	quantized	GGUF	1.54B	S	apache-2.0	Deeps03	3	3.9k
featherless-ai-quants/Qwen-Qwen2-1.5B-GGUF	quantized	GGUF	—	—	—	featherless-ai-quants	0	2.6k
oxyapi/oxy-1-micro	finetune	—	1.54B	S	apache-2.0	oxyapi	11	1.7k
arcee-ai/WitchLM-1.5B	finetune	—	1.54B	S	apache-2.0	arcee-ai	7	1.7k
DevQuasar-8/Qwen.Qwen2-1.5B-GGUF	quantized	GGUF	—	—	—	DevQuasar-8	0	1.8k
mlx-community/dolphin-2.9.3-qwen2-1.5b-8bit	finetune	-8bit	434M	S	apache-2.0	mlx-community	3	977
FreedomIntelligence/Apollo2-1.5B	finetune	—	1.78B	S	apache-2.0	FreedomIntelligence	1	891
nischay185/konkani-qwen2-1.5b	adapter	—	1.54B	S	apache-2.0	nischay185	0	777
kunal732/QwenReader	merge	—	1.54B	S	—	kunal732	0	708
amd/Qwen2-1.5B-onnx-ryzenai-npu	quantized	—	—	—	mit	amd	1	648
mlx-community/dolphin-2.9.3-qwen2-1.5b-2bit	finetune	—	145M	S	apache-2.0	mlx-community	2	577
rombodawg/rombos_Replete-Coder-Qwen2-1.5b	finetune	—	1.54B	S	apache-2.0	rombodawg	2	565
femiari/Qwen2-1.5Moe	finetune	—	4.09B	M	apache-2.0	femiari	2	548
AksaraLLM/AksaraLLM-Qwen-1.5B	finetune	—	1.78B	S	apache-2.0	AksaraLLM	0	627
BEE-spoke-data/Qwen2-1.5B-stepbasin-books	finetune	—	1.54B	S	apache-2.0	BEE-spoke-data	1	574
TheBlueObserver/Qwen2.5-1.5B-Instruct-4bit-128-GPTQ-MLX-LinearOnly	quantized	-4bit	1.54B	S	apache-2.0	TheBlueObserver	0	585
macadeliccc/Samantha-Qwen2-1.5B	finetune	—	1.54B	S	—	macadeliccc	0	566
mlx-community/dolphin-2.9.3-qwen2-1.5b-4bit	finetune	-4bit	241M	S	apache-2.0	mlx-community	1	435
Flowersea37/Qwen-1.5b-sft-qlora	adapter	—	—	—	apache-2.0	Flowersea37	0	478
AksaraLLM/aksarallm-1.5b-v2-checkpoint	finetune	—	1.78B	S	apache-2.0	AksaraLLM	0	422
Satyach/results	finetune	—	1.78B	S	—	Satyach	0	419
drmcbride/Qwen2-1.5B-Q2_K-GGUF	quantized	Q2_K	—	—	apache-2.0	drmcbride	0	385
PrompTartLAB/qwen2_1.5B_PTT_en_ko	finetune	—	1.54B	S	—	PrompTartLAB	0	355
AksaraLLM/AksaraLLM-Qwen-1.5B-v5-public	finetune	—	1.78B	S	apache-2.0	AksaraLLM	0	347
amy011872/LawToken-1.5B-baseline	finetune	—	1.54B	S	apache-2.0	amy011872	0	305
EndLessTime/gating_network_qwen	finetune	—	1.54B	S	apache-2.0	EndLessTime	0	298

Repository changelog

5 commits

Repository changelog 5 commits

Date	Commit	SHA
2024-06-06	update README.md	8a16abf
2024-06-06	Update README.md	c6c2fee
2024-05-31	Update config.json	537f5ab
2024-05-31	Initial commit	2aa468f
2024-05-31	initial commit	aa2850d

All providers

1 provider · 1 endpoint

Where to run 1 providers · 1 endpoints

Hosted endpoints observed for Qwen2-1.5B, sorted by input price (USD per million tokens). Each row names its observer and date; prices change daily, so check the source before quoting.

Provider	Model ref	Context	Quant	$/M in	$/M out	tok/s	TTFT	Tools	Status	Observed
Featherless hf	Qwen/Qwen2-1.5B	—	—	—	—	—	—	—	live	2026-10-25

Articles and coverage

7 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.

Official3

Qwen2 LLM Released

qwenlm.github.io2024-06-06officialauthorityHN front pageabout the Qwen2 familyHN 261 pts · 130 comments4,713 words

Alibaba releases Qwen2, a new series of open-source models featuring five sizes ranging from 0.5B to 72B parameters. The update applies Group Query Attention to all sizes and trains on data in 27 additional languages besides English and Chinese. Specific instruction-tuned versions like Qwen2-7B-Instruct and Qwen2-72B-Instruct support extended context lengths up to 128K tokens, while smaller models adopt the Apache 2.0 license and the largest retains the Qianwen License.written, reviewed, cited

Qwen2-57B-A14B disables tie embedding because large sparse embeddings consume a large proportion of total model parameters.
Instruction-tuned models require the Needle in a Haystack task to verify correct long-context understanding beyond perplexity evaluation.
The Qwen2-72B-Instruct model performs comparably to GPT-4 in safety tests across four categories of multilingual unsafe queries.

Qwen2-Math

qwenlm.github.io2024-08-08officialauthorityHN front pageabout the Qwen2 familyHN 128 pts · 38 comments5,833 words

Qwen2-Math is a series of specialized math language models built upon the Qwen2 LLMs that significantly outperforms the mathematical capabilities of open-source models and even closed-source models. The largest model, Qwen2-Math-72B-Instruct, is evaluated on benchmarks including GSM8K, Math, and MMLU-STEM using few-shot chain-of-thought prompting. Instruction tuning combines a dense reward signal with a binary signal via Rejection Sampling and reinforcement learning with Group Relative Policy Optimization (GRPO) after SFT.written, reviewed, cited

The model series primarily supports English, with bilingual English and Chinese math models scheduled for release soon.
Evaluation of multi-choice benchmarks including MMLU STEM and GaoKao problems uses a 5-shot setting instead of zero-shot.
Performance reporting for complex competition evaluations includes Greedy, Maj@64, RM@64, and RM@256 metrics.

An Analysis of Chinese LLM Censorship and Bias with Qwen 2 Instruct

huggingface.co2024-06-09authorityabout the Qwen2 familyHN 16 pts · 6 comments3,368 words

Qwen 2 Instruct models operate under an Apache 2.0 license but exhibit unique government-mandated alignment with significantly (>80%) fewer refusals in Chinese than English. The deccp dataset documents these behaviors, showing that responses to sensitive topics often include defensive nationalist propaganda or warnings rather than standard refusals. While the models are competitive on benchmarks, their output varies wildly in tone depending on the language used.written, reviewed, cited

Qwen2-72B-Instruct ranks just below Llama-3-70b-Instruct on the ChatArena Leaderboard several days after release.
The deccp dataset contains approximately 95 refusal questions used to evaluate Qwen2 Instructs censorship behavior.

Analysis3

Puro-2B: Poor Lab's Qwen2-1.5B Trained on RTX 5090 within $5090

arxiv.org2026-09-01authorityHN 3 pts · 0 comments485 words

Poor Lab releases Puro-2B, an open pretraining recipe that trains Qwen2-1.5B models from scratch on up to 1.4 trillion tokens with FP8 precision on consumer-grade RTX 5090 GPUs. The best model approaches Qwen2.5-1.5B performance under the evaluation protocol with a compute cost of less than $6.9K. A derived Puro Cost Scaling Law suggests that about $4.4K is sufficient to reach the performance of Qwen2-1.5B. The full training recipe, including data, code, and model weights, is available under Apache 2.0.written, reviewed, cited

Alibaba Qwen2.5-Omni-7B: Open Source End-to-End Multimodal AI Model

alizila.com2025-03-27about the Qwen2 familyHN 29 pts · 3 comments

Qwen2.5-Max: The Next Leap in AI Innovation

medium.com2025-01-29about the Qwen2 familyHN 17 pts · 4 comments

All coverage tracked for this model (7) — append-only, nothing is ever removed

Title	Source	Published	First seen	Signals
Puro-2B: Poor Lab's Qwen2-1.5B Trained on RTX 5090 within $5090	arxiv.org	2026-09-01	2026-10-07	hn
Alibaba Qwen2.5-Omni-7B: Open Source End-to-End Multimodal AI Model	alizila.com	2025-03-27	2026-10-07	hn
Qwen2.5-Max: The Next Leap in AI Innovation	medium.com	2025-01-29	2026-10-07	hn
Llama.cpp Now Supports Qwen2-VL (Vision Language Model)	github.com	2024-12-14	2026-10-07	hn
Qwen2-Math	qwenlm.github.io official	2024-08-08	2026-10-07	hn
An Analysis of Chinese LLM Censorship and Bias with Qwen 2 Instruct	huggingface.co	2024-06-09	2026-10-07	hn
Qwen2 LLM Released	qwenlm.github.io official	2024-06-06	2026-10-07	hn

Similar models

3 models

Similar models

Qwen2.5-1.5B-Instruct · 1.54B
same size band (S), same task (text-generation), permissive license
Qwen2.5-1.5B · 1.54B
same size band (S), same task (text-generation), permissive license
Qwen2.5-Coder-1.5B-Instruct · 1.54B
same size band (S), same task (text-generation), permissive license

Sources

3 sources

Sources

Hub snapshot: 2026-10-25 · HF API: 2026-10-24 21:39 UTC · Providers: 2026-10-24 21:44 UTC · Articles: 2026-10-24 21:46 UTC · Summary: 2026-10-07 15:10 UTC · Last verified 2026-10-25

Follow Qwen2-1.5B

Watch this model

/feeds/models/Qwen--Qwen2-1.5B.xml(RSS 2.0) · every change to Qwen2-1.5B
/feeds/families/Qwen--Qwen2.xml(RSS 2.0) · every change across the Qwen2 family

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