Up to date
Qwen tongyi-qianwen-license-agreement out 3 years ago
72.3B parameters, text in, text out, 32K tokens of context.
Verified 2026-10-25 Rank #407 in LLMs · score 79.3 method v1 Sources 3 Downloads 30d 4.7k · #548 of 619 tracked
Picked by rule from 43 tracked pieces: the lab's own post, the most shared, a field report, the latest.
Strata runs the 125-billion-parameter Qwen3.8-Flash-Next model on consumer hardware under an MIT license with no cloud dependencies.
The 2 picks above, plus these 23. Top is the radar's score: discussion, source authority and the lab's own posts, fading with age.
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.
No API provider serves it, so its downloads are people running it themselves: 2 quantized builds and 5 Spaces.
Top Spaces running it: Justinrune/LLaMA-Factory ♥ 12 · CATIE-AQ/Guide_Evaluation_LLM ♥ 5 · officialhimanshu595/llama-factory ♥ 4
Weights only, an estimate: leave room for the KV cache. Method
About 43 GB at INT4, 87 GB at FP8, 174 GB at BF16: weights only, so leave room for the KV cache.
Estimate params × bytes per param × 1.2 · how we estimate
Weights, every dated article, and repo commits on one rail.
Teal is the weights (2023-11-26), amber triangles are 10 dated articles (each links to it), grey ticks are repo commits.
3 derivatives on the Hub. Which build to run, by runtime: the most downloaded in each class.
From the Hugging Face model tree; quantized builds in teal, fine-tunes, merges and adapters in greys.
Everything else, folded: the full tables, every source.
| Benchmark | Value | Setting / notes | Reported by | Date |
|---|---|---|---|---|
| author-reported AGIEval | 62.5 | 0-shot | model author | 2026-10-07 |
| author-reported Avg | 66.4 | — | model author | 2026-10-07 |
| author-reported BBH · Reason through hard problems | 67.7 | 3-shot | model author | 2026-10-07 |
| author-reported C-Eval · Work across languages | 83.3 | 5-shot | model author | 2026-10-07 |
| author-reported CMMLU · Work across languages | 83.6 | 5-shot | model author | 2026-10-07 |
| author-reported GSM8K · Reason through hard problems | 78.9 | 8-shot | model author | 2026-10-07 |
| author-reported GaokaoBench | 87.6 | 0-shot | model author | 2026-10-07 |
| author-reported HumanEval · Write code with agents | 35.4 | 0-shot | model author | 2026-10-07 |
| author-reported MATH-500 · Reason through hard problems | 35.2 | 4-shot | model author | 2026-10-07 |
| author-reported MBPP · Write code with agents | 52.2 | 3-shot | model author | 2026-10-07 |
| author-reported MMLU · Reason through hard problems | 77.4 | 5-shot | model author | 2026-10-07 |
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.
Listed in LLMs
| Pipeline tag | text-generation |
|---|---|
| Library | transformers |
| Architectures | QWenLMHeadModel |
| Model type | qwen |
| Params by dtype | BF16: 72,287,920,128 |
| Size band | XL |
| MoE | no |
| Context length | 32,768 (config.json max_position_embeddings) |
| Languages | zh, en |
| Papers | arXiv:2309.16609 |
| License | tongyi-qianwen-license-agreement (other) |
| Hub snapshot | 2026-10-25 |
vLLM: not in registry · QWenLMHeadModel · 7050ef881ff0835f6890a47db1859a24c65481d6SGLang: supported · QWenLMHeadModel · 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.
Weights-only memory for Qwen-72B: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.
| BF16 (2 bytes/param) | 174 GB · Multi-GPU node (<640 GB) |
|---|---|
| FP8 (1 byte/param) | 87 GB · Multi-GPU node (<640 GB) |
| INT4 (0.5 bytes/param) | 43 GB · Single high-end GPU (<80 GB) |
Weights only (params × bytes/param × 1.2 overhead); KV cache and activations extra.
finetune: 1quantized: 2
| Variant | Relation | Quant | Params | Band | License | Author | Likes | Downloads |
|---|---|---|---|---|---|---|---|---|
| Qwen/Qwen-72B-Chat | finetune | — | 72.3B | XL | other | Qwen | 163 | 154.7k |
| mradermacher/Qwen-72B-i1-GGUF | quantized | GGUF | — | — | other | mradermacher | 1 | 8.7k |
| mradermacher/Qwen-72B-GGUF | quantized | GGUF | — | — | other | mradermacher | 1 | 3.1k |
| Date | Commit | SHA |
|---|---|---|
| 2024-10-09 | Expose metadata link to next version of the model (#11) | b8e18ac |
| 2024-01-09 | Update README.md | 6dae9c8 |
| 2023-12-13 | update wechat | 6ccff2c |
| 2023-12-13 | update wechat | f19b2b9 |
| 2023-12-07 | update modeling_qwen.py | fec78c0 |
| 2023-12-06 | update modeling_qwen.py | 2380231 |
| 2023-12-04 | update modeling_qwen.py | f62c598 |
| 2023-12-03 | update modeling_qwen.py | 61cd91c |
| 2023-11-30 | update | 3c79efc |
| 2023-11-30 | update | 29606ad |
| 2023-11-30 | update | 4c539df |
| 2023-11-28 | update config, split in 2GB | 443804a |
| 2023-11-27 | Update config.json | b4a48ab |
| 2023-11-26 | init upload | 15a041f |
| 2023-11-26 | initial commit | b20608b |
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.
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).
Hub snapshot: 2026-10-25 · HF API: 2026-10-24 21:37 UTC · Providers: 2026-10-24 21:44 UTC · Articles: 2026-10-24 21:46 UTC · Summary: 2026-10-07 15:05 UTC · Last verified 2026-10-25
Email alerts are not open yet; the feeds above are the same data. See feeds for the item schema.
