Language Models · Directory rank #9
Qwen3.8-27B
One model that sees and reads.
Watch the reel
Qwen3.8-27B is a native vision-language model from Qwen with 27B parameters. It understands images, video and text together — diagrams, documents, even hour-long videos.
It has a 262,144 token context window natively, extendable to 1M, with flexible thinking control. It works with Transformers, vLLM and SGLang, under Apache 2.0. It is a big 27B model, so check your hardware first.
Key facts
- Native vision-language model, 27B parameters
- Understands images, video and text together
- Reads diagrams, documents and hour-long videos
- 262,144 token context native, up to 1M
- Flexible thinking control
- Works with Transformers, vLLM, SGLang; Apache 2.0
PikyAI has not tested it live. This page is a model card overview from the PikyAI reel.
PikyAI reel on YouTube. Watch on Instagram
How to use it
- Hugging Face par model page kholo aur hardware check karo — 27B model bara hai.
- Transformers, vLLM ya SGLang se model load karo.
- Image, video ya text input do — diagram, document ya lambi video bhi.
Pros and cons
Pros
- Images, video aur text ek hi model mein.
- 262k token window, 1M tak extend hoti hai.
- Thinking control apne kaam ke hisaab se adjust karo.
- Apache 2.0, open weights.
Cons
- Bara 27B model — hardware pehle check karo.
- Benchmarks maintainer-reported hain.
- PikyAI ne live test nahi kiya.
Alternative
- Qwen3-Coder — Qwen3-Coder coding ke liye hai — vision nahi, lekin code tasks ke liye ek option hai.
- MiniCPM-OCR — MiniCPM-OCR documents se text, tables aur formulas parhta hai — sirf OCR chahiye to ek halka option hai.
How this rank works
Directory rank is set by bid, not by editorial opinion. New listings start at $5, a higher bid outranks a lower one, and taking #1 costs at least $5 more than the current top bid. Paid ranks always carry a Sponsored label, and payments are final and non-refundable. This listing sits at the $5 floor as a PikyAI house listing, not a paid rank.