Up to date
Qwen Apache 2.0 out 12 months ago
33.4B parameters, images and text in, text out, 256K tokens of context. 1 provider serves it.
Verified 2026-10-25 Rank #145 in LLMs · score 93.5 method v1 Sources 4 Downloads 30d 195.7k · #252 of 619 tracked
Changed this week: 2 new quantized builds on the Hub (51 in all).2026-10-24
Picked by rule from 19 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 4 picks above, plus these 15. Top is the radar's score: discussion, source authority and the lab's own posts, fading with age.
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 |
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; heavily run locally (51 quantized builds); a big base for builders (93 fine-tunes, adapters and merges).
Top Spaces running it: nvidia/Cosmos3-Action-Viewer ♥ 46 · dmorawiec/Qwen-VL-Object-Detection ♥ 18 · Smalleast23/AI_Personal_ArtDirector_Pro ♥ 16 · Slait/Caption_Qwen-vl ♥ 2
Weights only, an estimate: leave room for the KV cache. Method
About 20 GB at INT4, 40 GB at FP8, 80 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.
144 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: 33,357,390,064 |
| Size band | L |
| 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-32B-Instruct: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.
| BF16 (2 bytes/param) | 80 GB · Multi-GPU node (<640 GB) |
|---|---|
| FP8 (1 byte/param) | 40 GB · Single high-end GPU (<80 GB) |
| INT4 (0.5 bytes/param) | 20 GB · Single GPU (<24 GB) |
Weights only (params × bytes/param × 1.2 overhead); KV cache and activations extra.
quantized: 51finetune: 74adapter: 18merge: 1
| Variant | Relation | Quant | Params | Band | License | Author | Likes | Downloads |
|---|---|---|---|---|---|---|---|---|
| Qwen/Qwen3-VL-32B-Instruct-FP8 | quantized | -FP8 | 33.4B | L | apache-2.0 | Qwen | 61 | 4.8M |
| QuantTrio/Qwen3-VL-32B-Instruct-AWQ | quantized | AWQ | 33.4B | L | apache-2.0 | QuantTrio | 12 | 1.7M |
| unsloth/Qwen3-VL-32B-Instruct-bnb-4bit | quantized | -bnb | 33.4B | L | apache-2.0 | unsloth | 7 | 590.7k |
| huihui-ai/Huihui-Qwen3-VL-32B-Instruct-abliterated | finetune | — | 33.4B | L | apache-2.0 | huihui-ai | 31 | 303.9k |
| unsloth/Qwen3-VL-32B-Instruct-GGUF | quantized | GGUF | — | — | apache-2.0 | unsloth | 30 | 286.7k |
| Qwen/Qwen3-VL-32B-Instruct-GGUF | quantized | GGUF | — | — | apache-2.0 | Qwen | 28 | 210.4k |
| coder3101/Qwen3-VL-32B-Instruct-heretic-v2 | finetune | — | 33.4B | L | apache-2.0 | coder3101 | 11 | 76.5k |
| cyankiwi/Qwen3-VL-32B-Instruct-AWQ-4bit | quantized | AWQ | 7.03B | M | apache-2.0 | cyankiwi | 7 | 70.4k |
| RedHatAI/Qwen3-VL-32B-Instruct-NVFP4 | quantized | NVFP4 | 19.7B | L | apache-2.0 | RedHatAI | 8 | 58.2k |
| unsloth/Qwen3-VL-32B-Instruct-unsloth-bnb-4bit | quantized | -bnb | 33.4B | L | apache-2.0 | unsloth | 2 | 57.9k |
| pottokao/MiniMax-H3-TextEncoder-Qwen3VL-32B-abliterated-GGUF | quantized | GGUF | — | — | apache-2.0 | pottokao | 11 | 56.5k |
| nvidia/Cosmos-Reason2-32B | finetune | — | 33.4B | L | other | nvidia | 17 | 53.1k |
| lmstudio-community/Qwen3-VL-32B-Instruct-GGUF | quantized | GGUF | — | — | apache-2.0 | lmstudio-community | 5 | 41.2k |
| bartowski/Qwen_Qwen3-VL-32B-Instruct-GGUF | quantized | GGUF | — | — | — | bartowski | 7 | 39.8k |
| cybermotaz/Qwen3-VL-32B-Instruct-NVFP4 | quantized | NVFP4 | 17.8B | L | other | cybermotaz | 5 | 39k |
| unsloth/Qwen3-VL-32B-Instruct-1M-GGUF | quantized | GGUF | — | — | apache-2.0 | unsloth | 5 | 24.6k |
| unsloth/Qwen3-VL-32B-Instruct | finetune | — | 33.4B | L | apache-2.0 | unsloth | 5 | 19.7k |
| mradermacher/Qwen3-VL-32B-Instruct-GGUF | quantized | GGUF | — | — | apache-2.0 | mradermacher | 1 | 11.9k |
| lmstudio-community/Qwen3-VL-32B-Instruct-MLX-4bit | quantized | MLX | — | — | apache-2.0 | lmstudio-community | 1 | 11.9k |
| lmstudio-community/Qwen3-VL-32B-Instruct-MLX-8bit | quantized | MLX | — | — | apache-2.0 | lmstudio-community | 0 | 9.5k |
| CodeGoat24/UnifiedReward-2.0-qwen3vl-32b | finetune | — | 1.14M | S | mit | CodeGoat24 | 2 | 8.8k |
| mradermacher/Qwen3-VL-32B-Instruct-i1-GGUF | quantized | GGUF | — | — | apache-2.0 | mradermacher | 1 | 8.8k |
| RedHatAI/Qwen3-VL-32B-Instruct-FP8-dynamic | quantized | -FP8 | 33.4B | L | apache-2.0 | RedHatAI | 1 | 8.4k |
| lmstudio-community/Qwen3-VL-32B-Instruct-MLX-6bit | quantized | MLX | — | — | apache-2.0 | lmstudio-community | 0 | 6.4k |
| llmfan46/Qwen3-VL-32B-Instruct-ultra-uncensored-heretic | finetune | — | 33.4B | L | apache-2.0 | llmfan46 | 30 | 3.8k |
| cyankiwi/Qwen3-VL-32B-Instruct-AWQ-8bit | quantized | AWQ | 10.9B | L | apache-2.0 | cyankiwi | 3 | 5k |
| fbjr/qwen3-vl-32b-W4A16-AWQ-H3 | finetune | -W4A16 | — | — | apache-2.0 | fbjr | 3 | 5k |
| yairpatch/Qwen3-VL-32B-Instruct-GGUF | finetune | GGUF | — | — | apache-2.0 | yairpatch | 7 | 4.2k |
| lmstudio-community/Qwen3-VL-32B-Instruct-MLX-5bit | quantized | MLX | — | — | apache-2.0 | lmstudio-community | 0 | 4.4k |
| ThalisAI/Qwen3-VL-32B-Instruct-heretic | quantized | GGUF | — | — | other | ThalisAI | 0 | 4.4k |
| microsoft/AesCode-32B | finetune | — | 33.4B | L | apache-2.0 | microsoft | 55 | 1.4k |
| prithivMLmods/Qwen3-VL-32B-Instruct-abliterated-v1 | finetune | — | 33.4B | L | apache-2.0 | prithivMLmods | 10 | 3.4k |
| furiosa-ai/Qwen3-VL-32B-Instruct | finetune | — | 33.4B | L | apache-2.0 | furiosa-ai | 0 | 3.8k |
| lucataco/Qwen3-VL-32B-Instruct-Q4_K_M-GGUF | quantized | GGUF | — | — | apache-2.0 | lucataco | 4 | 3k |
| redponike/Qwen3-VL-32B-Instruct-GGUF | quantized | GGUF | — | — | apache-2.0 | redponike | 0 | 3.1k |
| bullerwins/Qwen3-VL-32B-Instruct-GGUF | quantized | GGUF | — | — | apache-2.0 | bullerwins | 2 | 2.8k |
| coder3101/Qwen3-VL-32B-Instruct-Heretic | finetune | — | 33.4B | L | apache-2.0 | coder3101 | 7 | 1.8k |
| Raretutor/Qwen3-VL-32B-MiniMax-H3-GGUF | quantized | GGUF | — | — | apache-2.0 | Raretutor | 0 | 2k |
| HiTZ/Latxa-Qwen3-VL-32B-Instruct | finetune | — | 1.14M | S | apache-2.0 | HiTZ | 2 | 1.7k |
| John1604/Qwen3-VL-32B-Instruct-gguf | quantized | gguf | — | — | apache-2.0 | John1604 | 0 | 1.6k |
Hosted endpoints observed for Qwen3-VL-32B-Instruct, 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/Qwen3-VL-32B-Instruct | — | — | — | — | — | — | — | live | 2026-10-25 |
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 21:35 UTC · Providers: 2026-10-24 21:44 UTC · Articles: 2026-10-24 21:45 UTC · Summary: 2026-10-07 15:02 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.
