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 › Qwen2 family

Qwen2-7B

Hugging Face ↗ Follow changes

Qwen Apache 2.0 out 2 years ago

7.62B parameters, text in, text out, 128K tokens of context. 1 provider serves it.

Verified 2026-10-25 Rank #400 in LLMs · score 79.6 method v1 Sources 3 Downloads 30d 223.4k · #242 of 619 tracked

Can I run it?4.6GB at INT4Estimatefits a 16 GB cardmethod
Can I use it?Apache 2.0Permissive; patent grantlicence file2026-10-24
Do people use it?223.4kdownloads, 30 days#242 of 619 we track · top 39%+2 likes, downloads +7% this week179 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.

Latest · 2 years ago

Qwen2-Math

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 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.
qwenlm.github.io 2024-08-08 128 HN pts · 38 comments 5.8k words about the Qwen2 family family

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.

  1. ArticleQwen2 LLM Releasedqwenlm.github.io · 2024-06-06 · 261 HN pts · official
  2. CodeLlama.cpp Now Supports Qwen2-VL (Vision Language Model)github.com · 2024-12-14
  3. ArticleAn Analysis of Chinese LLM Censorship and Bias with Qwen 2 Instructhuggingface.co · 2024-06-09 · 16 HN pts
  4. ArticleQwen2.5-Max: The Next Leap in AI Innovationmedium.com · 2025-01-29 · 17 HN pts
  5. ArticleQwen2-7B-Instruct with TensorRT-LLM: consistently high tokens/SECinferless.com · 2024-09-05 · 1 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 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; 46 quantized builds for local use; a big base for builders (315 fine-tunes, adapters and merges).

  • As a service1Featherless
  • On own hardware46quantized builds
  • Built on31585 fine-tunes · 225 adapters · 5 merges
  • In apps49Hugging Face Spaces

Top Spaces running it: ByteDance-Seed/BAGEL ♥ 222 · CZLC/BenCzechMark ♥ 26 · sanaka87/BAGEL-RecA ♥ 19 · KingNish/Bagel-7B-Demo ♥ 18

Will it fit?

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

Fits a 16 GB card at INT4

About 4.6 GB at INT4, 9.1 GB at FP8, 18 GB at BF16: weights only, so leave room for the KV cache.

16 GB card64 GB Mac2×80 GB8×80 GBINT4: about 4.6 GB of weights (estimate; KV cache and activations extra)INT44.6 GBFP8: about 9.1 GB of weights (estimate; KV cache and activations extra)FP89.1 GBBF16: about 18 GB of weights (estimate; KV cache and activations extra)BF1618 GB
16 GB card64 GB Mac2×80 GB8×80 GBINT4: about 4.6 GB of weights (estimate; KV cache and activations extra)INT44.6 GBFP8: about 9.1 GB of weights (estimate; KV cache and activations extra)FP89.1 GBBF16: about 18 GB of weights (estimate; KV cache and activations extra)BF1618 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.

6 dated articles and 4 repo commits so far

Teal is the weights (2024-06-04), amber triangles are 6 dated articles (each links to it), grey ticks are repo commits.

JulAugSepOctNovDecJanFebMarAprMayJunJulAugSepOctNovDecJanFebMarAprMayJunJulAugSepOct2024-06-06: 2 repo commits2024-06-04: 2 repo commitsQwen2 LLM Released · 2024-06-06Qwen2-Math · 2024-08-08An Analysis of Chinese LLM Censorship and Bias with Qwen 2 Instruct · 2024-06-09Alibaba Qwen2.5-Omni-7B: Open Source End-to-End Multimodal AI Model · 2025-03-27Qwen2.5-Max: The Next Leap in AI Innovation · 2025-01-29Qwen2-7B-Instruct with TensorRT-LLM: consistently high tokens/SEC · 2024-09-05Weights 2024-06-04 · 2024-06-04Weights 2024-06-04now · 1 provider
JulAugSepOctNovDecJanFebMarAprMayJunJulAugSepOctNovDecJanFebMarAprMayJunJulAugSepOct2024-06-06: 2 repo commits2024-06-04: 2 repo commitsQwen2 LLM Released · 2024-06-06Qwen2-Math · 2024-08-08An Analysis of Chinese LLM Censorship and Bias with Qwen 2 Instruct · 2024-06-09Alibaba Qwen2.5-Omni-7B: Open Source End-to-End Multimodal AI Model · 2025-03-27Qwen2.5-Max: The Next Leap in AI Innovation · 2025-01-29Qwen2-7B-Instruct with TensorRT-LLM: consistently high tokens/SEC · 2024-09-05Weights 2024-06-04 · 2024-06-04Weights 2024-06-04now · 1 provider

What people built on it

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

361 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.

46 quantized (Hugging Face model tree)85 fine-tunes (Hugging Face model tree)5 merges (Hugging Face model tree)225 adapters (Hugging Face model tree)46 quantized85 fine-tunes5 merges225 adapters
46 quantized (Hugging Face model tree)85 fine-tunes (Hugging Face model tree)5 merges (Hugging Face model tree)225 adapters (Hugging Face model tree)46 quantized85 fine-tunes5 merges225 adapters
GGUF legraphista/Qwen2-7B-IMat-GGUF for llama.cpp, Ollama, LM Studio 54.3k downloads MLX mlx-community/Molmo-7B-D-0924-bf16 for Apple silicon 7.3k downloads

All variants →

Reference

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

Benchmarks

21 rows

Benchmarks 21 rows

measured third-party leaderboard rowauthor-reported number from the model card or eval files (21)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.
BenchmarkValueSetting / notesReported byDate
author-reported ARC Challenge · Chat and follow instructions60.6—model author2026-10-07
author-reported BBH · Reason through hard problems62.6—model author2026-10-07
author-reported C-Eval · Work across languages83.2—model author2026-10-07
author-reported CMMLU · Work across languages83.9—model author2026-10-07
author-reported GPQA Diamond · Reason through hard problems31.8—model author2026-10-07
author-reported GSM8K · Reason through hard problems79.9—model author2026-10-07
author-reported HellaSwag · Chat and follow instructions80.7—model author2026-10-07
author-reported HumanEval · Write code with agents51.2—model author2026-10-07
author-reported HumanEval · Write code with agents54.2—model author2026-10-07
author-reported MATH-500 · Reason through hard problems44.2—model author2026-10-07
author-reported MATH-500 · Reason through hard problems57.5—model author2026-10-07
author-reported MBPP · Write code with agents65.9—model author2026-10-07
author-reported MMLU · Reason through hard problems70.3—model author2026-10-07
author-reported MMLU-Pro · Reason through hard problems40—model author2026-10-07
author-reported Multi-Exam59.2—model author2026-10-07
author-reported Multi-Translation31.5—model author2026-10-07
author-reported Multi-Understanding72—model author2026-10-07
author-reported MultiPL-E · Write code with agents46.3—model author2026-10-07
author-reported Theorem QA31.1—model author2026-10-07
author-reported TruthfulQA · Chat and follow instructions54.2—model author2026-10-07
author-reported Winogrande · Chat and follow instructions77—model author2026-10-07

What the card says

1 cited sentence

Summary generated, cited

The model uses a Transformer architecture with SwiGLU activation and group query attention.[1]

  1. “Transformer architecture with SwiGLU activation, attentio…”

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

Pipeline tagtext-generation
Librarytransformers
ArchitecturesQwen2ForCausalLM
Model typeqwen2
Dtypebfloat16
Params by dtypeBF16: 7,615,616,512
Size bandM
MoEno
Context length131,072 (config.json max_position_embeddings)
Languagesen
LicenseApache 2.0 (apache-2.0)
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 Qwen2-7B: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.

BF16 (2 bytes/param)18 GB · Single GPU (<24 GB)
FP8 (1 byte/param)9.1 GB · Single GPU (<24 GB)
INT4 (0.5 bytes/param)4.6 GB · Single GPU (<24 GB)

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

All variants

40 variants

Lineage

Original model Family: Qwen/Qwen2 361 derivatives on the Hub

finetune: 85quantized: 46adapter: 225merge: 5

Top variants (40 shown, ranked by downloads)

Relation
Quant
Size
40 rows
Variant Relation Quant Params Band License Author Likes Downloads
MaziyarPanahi/calme-2.3-qwen2-7b finetune — 7.62B M apache-2.0 MaziyarPanahi 2 175.9k
MaziyarPanahi/calme-2.4-qwen2-7b finetune — 7.62B M apache-2.0 MaziyarPanahi 1 175.8k
MaziyarPanahi/calme-2.6-qwen2-7b finetune — 7.62B M apache-2.0 MaziyarPanahi 2 175.6k
MaziyarPanahi/calme-2.5-qwen2-7b finetune — 7.62B M apache-2.0 MaziyarPanahi 1 175.6k
MaziyarPanahi/calme-2.7-qwen2-7b finetune — 7.62B M apache-2.0 MaziyarPanahi 2 175.3k
MaziyarPanahi/calme-2.8-qwen2-7b finetune — 7.62B M apache-2.0 MaziyarPanahi 6 149.2k
MaziyarPanahi/calme-2.2-qwen2-7b finetune — 7.62B M apache-2.0 MaziyarPanahi 1 142.2k
MaziyarPanahi/calme-2.1-qwen2-7b finetune — 7.62B M apache-2.0 MaziyarPanahi 1 135.2k
dphn/dolphin-2.9.2-qwen2-7b finetune — 7.62B M apache-2.0 dphn 76 69.3k
legraphista/Qwen2-7B-IMat-GGUF quantized GGUF — — apache-2.0 legraphista 0 54.3k
Weyaxi/Einstein-v7-Qwen2-7B finetune — 7.62B M other Weyaxi 42 49.2k
macadeliccc/Samantha-Qwen-2-7B finetune — 7.62B M apache-2.0 macadeliccc 3 46.6k
bartowski/Einstein-v7-Qwen2-7B-GGUF quantized GGUF — — other bartowski 7 41.4k
Efficient-Large-Model/Qwen2-VL-7B-Instruct finetune — 8.29B M apache-2.0 Efficient-Large-Model 0 38.7k
ATH-MaaS/Marco-LLM-GLO finetune — 7.62B M apache-2.0 ATH-MaaS 6 31.3k
bartowski/Samantha-Qwen-2-7B-GGUF quantized GGUF — — apache-2.0 bartowski 8 26.2k
FreedomIntelligence/Apollo2-7B finetune — 7.62B M apache-2.0 FreedomIntelligence 3 22.3k
mradermacher/Qwen2-7B-i1-GGUF quantized GGUF — — apache-2.0 mradermacher 0 13.2k
mradermacher/Qwen2-7B-GGUF quantized GGUF — — apache-2.0 mradermacher 0 13.1k
QuantFactory/Qwen2-7B-GGUF quantized GGUF — — apache-2.0 QuantFactory 3 11.4k
fedric95/Qwen2-7B-GGUF quantized GGUF — — apache-2.0 fedric95 0 11.2k
mlx-community/Molmo-7B-D-0924-bf16 finetune -bf16 8.02B M apache-2.0 mlx-community 1 7.3k
QuantFactory/calme-2.8-qwen2-7b-GGUF quantized GGUF — — apache-2.0 QuantFactory 1 6.3k
QuantFactory/Apollo2-7B-GGUF quantized GGUF — — apache-2.0 QuantFactory 3 4.8k
Grogros/Qwen2-7B-OurSafecoder finetune — 7.62B M apache-2.0 Grogros 0 4.8k
Jerry-Adams/Qwen-Qwen2-7B-1728028316 adapter — — — — Jerry-Adams 0 4.6k
oumi-ai/Molmo-7B-D-0924 finetune — 8.02B M apache-2.0 oumi-ai 0 3.7k
QuantFactory/Replete-LLM-Qwen2-7b-GGUF quantized GGUF — — apache-2.0 QuantFactory 1 3.5k
mlx-community/Molmo-7B-D-0924-4bit finetune -4bit 8.02B M apache-2.0 mlx-community 2 2.7k
Edens-Gate/Henbane-7b-attempt2 finetune — — — apache-2.0 Edens-Gate 1 2.3k
tsinghua-ee/video-SALMONN-2 finetune — 8.72B M apache-2.0 tsinghua-ee 4 2k
polyverse/Qwen2-7B-Q4_K_M-GGUF quantized GGUF — — apache-2.0 polyverse 0 1.9k
FreedomIntelligence/Apollo-MoE-7B finetune — 10.1B L apache-2.0 FreedomIntelligence 10 1.4k
NikolayKozloff/Qwen2-7B-Q4_K_M-GGUF quantized GGUF — — apache-2.0 NikolayKozloff 1 1.7k
NikolayKozloff/Qwen2-7B-Q6_K-GGUF quantized Q6_K — — apache-2.0 NikolayKozloff 1 1.6k
amy011872/LawToken-7B-a2 finetune — 7.71B M apache-2.0 amy011872 0 1.5k
featherless-ai-quants/Qwen-Qwen2-7B-GGUF quantized GGUF — — — featherless-ai-quants 0 1.4k
tanliboy/zephyr-qwen2-7b-sft finetune — 7.62B M apache-2.0 tanliboy 0 1.3k
mlx-community/dolphin-2.9.2-qwen2-7b-4bit finetune -4bit 7.62B M apache-2.0 mlx-community 3 987
NikolayKozloff/Qwen2-7B-Q8_0-GGUF quantized Q8_0 — — apache-2.0 NikolayKozloff 2 976

Repository changelog

4 commits

Repository changelog 4 commits

DateCommitSHA
2024-06-06 update README.md 453ed15
2024-06-06 Update README.md 0546a58
2024-06-04 init upload 9b039ab
2024-06-04 initial commit f5f7982

All providers

1 provider · 1 endpoint

Where to run 1 providers · 1 endpoints

Hosted endpoints observed for Qwen2-7B, 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/Qwen2-7B — — — — — — — 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

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
Qwen2-7B-Instruct with TensorRT-LLM: consistently high tokens/SEC
inferless.com2024-09-05HN 1 pts · 1 comments913 words
All coverage tracked for this model (7) — append-only, nothing is ever removed
TitleSourcePublishedFirst seenSignals
Alibaba Qwen2.5-Omni-7B: Open Source End-to-End Multimodal AI Modelalizila.com2025-03-272026-10-07hn
Qwen2.5-Max: The Next Leap in AI Innovationmedium.com2025-01-292026-10-07hn
Llama.cpp Now Supports Qwen2-VL (Vision Language Model)github.com2024-12-142026-10-07hn
Qwen2-7B-Instruct with TensorRT-LLM: consistently high tokens/SECinferless.com2024-09-052026-10-07hn
Qwen2-Mathqwenlm.github.io official2024-08-082026-10-07hn
An Analysis of Chinese LLM Censorship and Bias with Qwen 2 Instructhuggingface.co2024-06-092026-10-07hn
Qwen2 LLM Releasedqwenlm.github.io official2024-06-062026-10-07hn

Similar models

3 models

Similar models

  • Llama-3.1-8B-Instruct · 8.03B
    same size band (M), same task (text-generation), more providers
  • Qwen2.5-7B-Instruct · 7.62B
    same size band (M), same task (text-generation), more providers, permissive license
  • Qwen3-8B · 8.19B
    same size band (M), same task (text-generation), more providers, 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 Qwen2-7B

Watch this model

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