Qwen3-Embedding-8B vs Qwen3-VL-Embedding-8B

Qwen3-Embedding-8B

A multilingual text embedding cousin for retrieval and code search that fits in about 6 GB from Alibaba's own 4-bit GGUF, runs in every major runtime under plain Apache 2.0 and has several hosts. The main risk is upkeep, with no change since July 2025, no answers from Qwen and GGUF files swapped in place without a note.

Qwen3-VL-Embedding-8B

A multimodal embedding cousin that puts text, images, screenshots and video in one vector space for visual document and video retrieval, under plain Apache 2.0. Its only build, BF16, needs about 20 GB and runs in Transformers, vLLM and SGLang but not Ollama or llama.cpp. The main risk is the publisher's script on the Transformers path.

Qwen3-Embedding-8B

1. Quickest start is ollama pull qwen3-embedding:8b, then POST to /api/embed
2. With llama.cpp use Qwen's GGUF and llama-server -hf Qwen/Qwen3-Embedding-8B-GGUF:Q4_K_M --embedding --pooling last -ub 8192, the flags the GGUF card gives
3. Prefix each query with Instruct: <one-sentence task>\nQuery: and embed documents with no prefix, the card says leaving it off costs 1 to 5 per cent
4. Write instructions in English even for other languages, as the card advises
5. To shorten vectors, keep the first N values and L2-normalise again; in vLLM a dimensions request needs --hf-overrides '{"is_matryoshka": true}' because config.json doesn't declare it
6. Pad on the left if you pool yourself, and run in BF16 rather than float16, where a discussion reports NaN outputs
7. Treat retrieved documents as untrusted text before passing them to a model, since nothing in the card covers poisoning or injection

Qwen3-VL-Embedding-8B

1. Simplest start is Sentence Transformers 5.4 or later, SentenceTransformer('Qwen/Qwen3-VL-Embedding-8B'), which applies the template and the default instruction and runs no repository code
2. To serve it, vllm serve Qwen/Qwen3-VL-Embedding-8B --runner pooling, then POST messages with a system instruction and image_url or text parts to /v1/embeddings; vLLM's docs say its video preprocessing differs slightly from Qwen's qwen_vl_utils
3. Put the task instruction in the system turn, and write it in English, as the card advises; the default is "Represent the user's input."
4. To shorten vectors, keep the first N of the 4,096 values and L2-normalise again; the card gives 64 as the floor, and the report says int8 vectors hold up while binary ones lose noticeably
5. For several images per input use the reference embedder in QwenLM/Qwen3-VL-Embedding's src, which is newer than the script in the Hugging Face repository
6. Keep inputs under 32K tokens; the reference embedder defaults to 8,192, 1,843,200 pixels per image and 64 frames per video
7. Treat retrieved screenshots and documents as untrusted, since text in an image can carry instructions to whatever model reads the results