Vision-language models trending today
12 of the 40 models on Hugging Face’s trending list carry this task tag today, in Hugging Face’s order, with entries held back when likes are not matched by downloads.
Collected · snapshot shipped with the site build, not a live feed
12 of 12 models · Hugging Face order
Number 1: deepseek-ai/DeepSeek-V4.1-Flash-1017
Image text to text763B parameters
No. 1 by Hugging Face trending score (2,224) · repository created 7 days ago · +18 likes since 2026-10-23.
57.7K downloads in 30 days · 2,617 likes
MIT· commercial use allowedNumber 2: Qwen/Qwen3.8-27B
Image text to text27.8B parameters
No. 10 by Hugging Face trending score (314) · +7 likes since 2026-10-23 · 7,710,011 downloads in the last 30 days.
7.71M downloads in 30 days · 18.3K likes
Apache-2.0· commercial use allowedNumber 3: dealignai/DeepSeek-V4.1-Flash-1017-UNCENSORED-FP8
Image text to text763B parameters
No. 13 by Hugging Face trending score (268) · repository created 6 days ago · +1 likes since 2026-10-23.
4,141 downloads in 30 days · 307 likes
MIT· commercial use allowedNumber 4: MiniMaxAI/MiniMax-M3.1-Flash
Image text to text22B parameters
No. 14 by Hugging Face trending score (262) · repository created 7 days ago · +12 likes since 2026-10-23.
1,096 downloads in 30 days · 268 likes
Other (custom terms)· read the termsNumber 5: orcarouter/OrcaCyber-Nano-1.0
Image text to text
No. 20 by Hugging Face trending score (199) · repository created 7 days ago · +4 likes since 2026-10-23.
22.3K downloads in 30 days · 214 likes
Other (custom terms)· read the termsNumber 6: meta-models/Muse-Spark-Open-816B-A42B
Image text to text822B parameters
No. 24 by Hugging Face trending score (171) · +1 likes since 2026-10-23 · 5,196 downloads in the last 30 days.
5,196 downloads in 30 days · 679 likes
Other (custom terms)· read the termsNumber 7: nvidia/Qwen3.8-Flash-Next-FP4-Local
Image text to text96.5B parameters
No. 26 by Hugging Face trending score (158) · 11,197 downloads in the last 30 days.
11.2K downloads in 30 days · 328 likes
Other (custom terms)· read the termsNumber 8: DavidAU/Qwen3.8-27B-TURBO-Fable-Cold-Fusion-735-882-Heretic-Uncensored-NEO-CODER-MAX-MTP-GGUF
Image text to text
No. 27 by Hugging Face trending score (157) · +3 likes since 2026-10-23 · 1,679,145 downloads in the last 30 days.
1.68M downloads in 30 days · 2,024 likes
Apache-2.0· commercial use allowedNumber 9: Qwen/Qwen3.8-Flash-Next
Image text to text18B parameters
No. 30 by Hugging Face trending score (145) · +3 likes since 2026-10-23 · 2,788,737 downloads in the last 30 days.
2.79M downloads in 30 days · 6,443 likes
Other (custom terms)· read the termsNumber 10: unsloth/DeepSeek-V4.1-Flash-1017-GGUF
Image text to text
No. 36 by Hugging Face trending score (130) · repository created 6 days ago.
29.2K downloads in 30 days · 135 likes
MIT· commercial use allowedNumber 11: HauhauCS/Qwen3.8-27B-Uncensored-HauhauCS-Aggressive-MTP-GGUF
Image text to text
No. 37 by Hugging Face trending score (127) · +2 likes since 2026-10-23 · 2,277,012 downloads in the last 30 days.
2.28M downloads in 30 days · 2,038 likes
Apache-2.0· commercial use allowedNumber 12: deepseek-ai/DeepSeek-V4.1-Flash
Image text to text763B parameters
No. 40 by Hugging Face trending score (121) · 1,478,503 downloads in the last 30 days.
1.48M downloads in 30 days · 4,716 likes
MIT· commercial use allowed
Positions follow Hugging Face’s trending order with 0 of 40 entries held back for likes that downloads do not confirm. Sorting by another column changes the order on your screen only.
Source: Hugging Face public API · updated
About vision-language models
Image-text-to-text is the Hugging Face tag for models that take images together with text and answer in text: document reading, chart and screenshot understanding, visual question answering. Most recent general-purpose chat models are published under this tag because they accept images, so this list overlaps heavily with what people call "LLMs".
Before you pick one
Check that the conversion kept the vision part
A GGUF conversion of a vision-language model needs a separate projector file to accept images. Without it the model works as text only. The repository file list tells you whether the projector is included.
Look at the input resolution
How many pixels the model sees per image decides whether it can read small print. It is stated in the model card, not in the API, and it differs between models of the same size.
Size still drives cost
Images are turned into hundreds or thousands of tokens each. A model that fits in memory for text can run out with several images in the context.
Questions readers ask
Is image-text-to-text the same as image generation?
No. These models read images and produce text. Models that produce images from a prompt are tagged text-to-image.
Why are well-known chat models listed here and not under text generation?
The uploader chooses one task tag per repository. When a chat model also accepts images, publishers usually pick image-text-to-text. Text-only fine-tunes and conversions of the same model may be tagged text-generation.
Can these models read PDFs?
They read page images. A PDF has to be rendered to images first, one per page, which your own code or tool does before calling the model.