Up to date
Qwen Apache 2.0 out 12 months ago
2.13B parameters, images and text in, text out, 256K tokens of context.
Verified 2026-10-25 Rank #637 in LLMs · score 73.3 method v1 Sources 3 Downloads 30d 56.9k · #373 of 619 tracked
Changed this week: 1 new quantized build on the Hub (21 in all).2026-10-24
Picked by rule from 10 tracked pieces: the lab's own post, the most shared, a field report, the latest.
OpenAI released gpt-oss-120b and gpt-oss-20b, their first open-weight models since GPT-2 in 2019.
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.
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: 21 quantized builds and 13 Spaces.
Top Spaces running it: baohuynhbk14/Qwen3-VL-Demo ♥ 57 · nvidia/Cosmos3-Action-Viewer ♥ 46 · Slait/Caption_Qwen-vl ♥ 2 · nvidia/Cosmos3-Action-Viewer-Prerelease ♥ 1
Weights only, an estimate: leave room for the KV cache. Method
About 1.3 GB at INT4, 2.6 GB at FP8, 5.1 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 (2025-10-19), amber triangles are 4 dated articles (each links to it), grey ticks are repo commits.
57 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.
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 | image-text-to-text |
|---|---|
| Library | transformers |
| Architectures | Qwen3VLForConditionalGeneration |
| Model type | qwen3_vl |
| Dtype | bfloat16 |
| Params by dtype | BF16: 2,127,532,032 |
| Size band | M |
| MoE | no |
| Context length | 262,144 (config.json max_position_embeddings) |
| Papers | arXiv:2505.09388, arXiv:2502.13923, arXiv:2409.12191, arXiv:2308.12966 |
| License | Apache 2.0 (apache-2.0) |
| Hub snapshot | 2026-10-25 |
vLLM: supported · Qwen3VLForConditionalGeneration · 7050ef881ff0835f6890a47db1859a24c65481d6SGLang: supported · Qwen3VLForConditionalGeneration · 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 Qwen3-VL-2B-Thinking: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.
| BF16 (2 bytes/param) | 5.1 GB · Single GPU (<24 GB) |
|---|---|
| FP8 (1 byte/param) | 2.6 GB · Edge (<4 GB) |
| INT4 (0.5 bytes/param) | 1.3 GB · Edge (<4 GB) |
Weights only (params × bytes/param × 1.2 overhead); KV cache and activations extra.
quantized: 21finetune: 25adapter: 11
| Variant | Relation | Quant | Params | Band | License | Author | Likes | Downloads |
|---|---|---|---|---|---|---|---|---|
| unsloth/Qwen3-VL-2B-Thinking-GGUF | quantized | GGUF | — | — | apache-2.0 | unsloth | 10 | 151k |
| unsloth/Qwen3-VL-2B-Thinking-1M-GGUF | quantized | GGUF | — | — | apache-2.0 | unsloth | 4 | 32.2k |
| Qwen/Qwen3-VL-2B-Thinking-GGUF | quantized | GGUF | — | — | apache-2.0 | Qwen | 30 | 23k |
| bartowski/Qwen_Qwen3-VL-2B-Thinking-GGUF | quantized | GGUF | — | — | — | bartowski | 2 | 11.3k |
| furiosa-ai/Qwen3-VL-2B-Thinking | finetune | — | 2.13B | M | apache-2.0 | furiosa-ai | 0 | 9.1k |
| zeromodels/qwen3-vl-2b-thinking | finetune | — | — | — | apache-2.0 | zeromodels | 0 | 7.9k |
| unsloth/Qwen3-VL-2B-Thinking-unsloth-bnb-4bit | quantized | -bnb | 2.17B | M | apache-2.0 | unsloth | 1 | 6.8k |
| lmstudio-community/Qwen3-VL-2B-Thinking-MLX-8bit | quantized | MLX | 2.13B | M | apache-2.0 | lmstudio-community | 1 | 6.6k |
| NexaAI/Qwen3-VL-2B-Thinking-GGUF | quantized | GGUF | — | — | — | NexaAI | 14 | 5.3k |
| mradermacher/Qwen3-VL-2B-Thinking-i1-GGUF | quantized | GGUF | — | — | apache-2.0 | mradermacher | 0 | 5.4k |
| lmstudio-community/Qwen3-VL-2B-Thinking-GGUF | quantized | GGUF | — | — | apache-2.0 | lmstudio-community | 0 | 4.2k |
| unsloth/Qwen3-VL-2B-Thinking | finetune | — | 2.13B | M | apache-2.0 | unsloth | 1 | 3.7k |
| mradermacher/Qwen3-VL-2B-Thinking-GGUF | quantized | GGUF | — | — | apache-2.0 | mradermacher | 0 | 2.8k |
| cyankiwi/Qwen3-VL-2B-Thinking-AWQ-4bit | quantized | AWQ | 2.13B | M | apache-2.0 | cyankiwi | 0 | 2.8k |
| happy8825/sft-20251124 | finetune | — | 2.13B | M | other | happy8825 | 0 | 2.7k |
| unsloth/Qwen3-VL-2B-Thinking-bnb-4bit | quantized | -bnb | 2.17B | M | apache-2.0 | unsloth | 0 | 2.2k |
| redponike/Qwen3-VL-2B-Thinking-GGUF | quantized | GGUF | — | — | apache-2.0 | redponike | 0 | 1.9k |
| happy8825/sft-20251126 | finetune | — | 2.13B | M | other | happy8825 | 0 | 1.5k |
| John1604/Qwen3-VL-2B-Thinking-gguf | quantized | gguf | — | — | apache-2.0 | John1604 | 0 | 1.3k |
| DevQuasar/Qwen.Qwen3-VL-2B-Thinking-GGUF | quantized | GGUF | — | — | — | DevQuasar | 0 | 1.3k |
| madrisight/MadriMed-VL-2B | finetune | — | 2.13B | M | apache-2.0 | madrisight | 3 | 808 |
| taobao-mnn/Qwen3-VL-2B-Thinking-MNN | quantized | — | — | — | apache-2.0 | taobao-mnn | 0 | 725 |
| coder3101/Qwen3-VL-2B-Thinking-heretic | finetune | — | 2.13B | M | apache-2.0 | coder3101 | 0 | 648 |
| prithivMLmods/Qwen3-VL-2B-Thinking-abliterated-v1 | finetune | — | 2.13B | M | apache-2.0 | prithivMLmods | 3 | 333 |
| Nhaass/Qwen3-VL-2B-ChartQA | adapter | — | 2.13B | M | apache-2.0 | Nhaass | 2 | 368 |
| remyxai/SpaceQwen3-VL-2B-Thinking | finetune | — | 2.13B | M | apache-2.0 | remyxai | 3 | 311 |
| cybermotaz/qwen3-vl-2b-thinking-nvfp4-w4a16 | quantized | nvfp4 | 1.51B | S | apache-2.0 | cybermotaz | 1 | 325 |
| madrisight/madrimed1.2-VL-2B | finetune | — | 2.44B | M | apache-2.0 | madrisight | 2 | 186 |
| taobao-mnn/Qwen3-VL-2B-Thinking-Eagle3 | finetune | — | 137M | S | apache-2.0 | taobao-mnn | 0 | 205 |
| AXONVERTEX-AI-RESEARCH/Qwen3-VL-2B-Thinking-Q8_0-GGUF | quantized | Q8_0 | — | — | apache-2.0 | AXONVERTEX-AI-RESEARCH | 0 | 202 |
| KiwiMate/KiwiMate-Mini-1.1 | quantized | GGUF | — | — | apache-2.0 | KiwiMate | 0 | 173 |
| cyankiwi/Qwen3-VL-2B-Thinking-AWQ-8bit | quantized | AWQ | 2.13B | M | apache-2.0 | cyankiwi | 0 | 169 |
| mistralrs-community/Qwen3-VL-2B-Thinking-UQFF | quantized | — | — | — | — | mistralrs-community | 0 | 165 |
| wabibito/Onyx-Qwen3VL-2B-S2V-Encoder | finetune | — | 2.13B | M | apache-2.0 | wabibito | 1 | 92 |
| adwel94/vision-safari-agent-lora | adapter | — | — | — | — | adwel94 | 0 | 135 |
| HumorR1/policy-qwen3vl-2b-grpo-newyorker | adapter | — | — | — | apache-2.0 | HumorR1 | 0 | 77 |
| adwel94/vision-emoji-recognition-lora | adapter | — | — | — | — | adwel94 | 0 | 55 |
| happy8825/sft-20251121 | finetune | — | 2.13B | M | other | happy8825 | 0 | 47 |
| HumorR1/policy-e1b-sft-thinking | finetune | — | 2.13B | M | apache-2.0 | HumorR1 | 0 | 44 |
| SeonghoonYu/Masking-KD | finetune | — | 2.44B | M | apache-2.0 | SeonghoonYu | 0 | 41 |
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.
| Title | Source | Published | First seen | Signals |
|---|---|---|---|---|
| Qwen3.5 122B and 35B models offer Sonnet 4.5 performance on local computers | venturebeat.com | 2026-02-28 | 2026-10-07 | hn |
| Qwen3-Coder-Next | qwen.ai official | 2026-02-03 | 2026-10-07 | hn |
| Qwen3-Max-Thinking | qwen.ai official | 2026-01-26 | 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-VL | qwen.ai official | 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 |
| 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 |
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 21:42 UTC · Providers: 2026-10-24 21:44 UTC · Articles: 2026-10-24 21:46 UTC · Summary: 2026-10-07 15:14 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.
