# Qwen3-VL-Embedding-2B by Qwen: Open-Weight Model Source: https://savrn.com/models/qwen3-vl-embedding-2b Markdown alternate of the page above; the site index is https://savrn.com/llms.txt --- ## Runs On What it takes to serve Qwen3-VL-Embedding-2B (2.1B parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them. | Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits | | --- | --- | --- | --- | --- | --- | | 16-bit | 4.3 GB | 5.1 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | | 8-bit | 2.1 GB | 2.6 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | | 4-bit | 1.1 GB | 1.3 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the [SAVRN Index](https://savrn.com/ai-index/pricing/gpus), read Oct 25, 2026. [Qwen3-VL-Embedding-2B on every accelerator the SAVRN Index prices, at every precision](https://savrn.com/models/qwen3-vl-embedding-2b/gpus) ## SAVRN's Notes on Qwen3-VL-Embedding-2B Retrieval over screenshots, video frames and text in one vector space is the job here, and at 2.1 billion parameters it is small enough to run beside the generator it feeds. The 16-bit footprint is 5.1 gigabytes of memory for 4.3 gigabytes of weights, so the cheapest listed setup, one MI300X with 192 gigabytes at $1.85 an hour, leaves most of the card free for a language model. The 262,144-token context is the number to watch: memory per request grows with what you feed in, so size the batch, not just the card. Apache 2.0 covers commercial use, modification and redistribution with the notices kept, which suits a customer-facing index. Two checks: the weights derive from Qwen3-VL-2B-Instruct, and the paper, arXiv:2601.04720, describes this embedder alongside a reranker the publisher pairs with it. The SAVRN Index lists no host prices yet, so the on-premises figure above is the only price today. SAVRN Research, 2026-09-18 ## Model Card By Qwen, published under apache-2.0, revision 9f2f7e710d6d. ### Highlights The Qwen3-VL-Embedding and Qwen3-VL-Reranker model series are the latest additions to the Qwen family, built upon the recently open-sourced and powerful Qwen3-VL foundation model. Specifically designed for multimodal information retrieval and cross-modal understanding, this suite accepts diverse inputs including text, images, screenshots, and videos, as well as inputs containing a mixture of these modalities. While the Embedding model generates high-dimensional vectors for broad applications like retrieval and clustering, the Reranker model is engineered to refine these results, establishing a comprehensive pipeline for state-of-the-art multimodal search. [Read the full model card (2,419 words)](https://savrn.com/models/qwen3-vl-embedding-2b/card) ## Configuration Architecture Qwen3VLForConditionalGeneration Context length (tokens) 262,144 Layers 28 Hidden size 2,048 Feed-forward size 6,144 Attention heads 16 Key/value heads 8 Head dimension 128 Vocabulary size 151,936 RoPE base 5,000,000 Model type qwen3_vl ## Identity and Version Repository Qwen/Qwen3-VL-Embedding-2B Publisher Qwen Task Sentence similarity Modality Text Library sentence-transformers Parameters 2.1B parameters Languages Not stated by the source Revision 9f2f7e710d6d81056aa5c0a4f04764fec6bb7bda First published 2026-01-07 Last updated 2026-04-16 ## Files and Weights 18 files, 4.3 GB in total. The weights are 1 file totalling 4.3 GB in safetensors. Weights1 file · 4.3 GB Configuration10 files · 19.3 KB Tokenizer4 files · 15.9 MB Documentation1 file · 24.2 KB Other1 file · 5.5 KB Repository1 file · 1.6 KB Every file | File | Type | Size | SHA-256 | | --- | --- | --- | --- | | model.safetensors | Weights | 4.3 GB | c73fa9caedde | | 1_Pooling/config.json | Configuration | 96 B | — | | added_tokens.json | Configuration | 707 B | — | | config.json | Configuration | 1.6 KB | — | | config_sentence_transformers.json | Configuration | 238 B | — | | modules.json | Configuration | 429 B | — | | preprocessor_config.json | Configuration | 783 B | — | | scripts/qwen3_vl_embedding.py | Configuration | 13.3 KB | — | | sentence_bert_config.json | Configuration | 770 B | — | | special_tokens_map.json | Configuration | 613 B | — | | video_preprocessor_config.json | Configuration | 817 B | — | | README.md | Documentation | 24.2 KB | — | | chat_template.jinja | Other | 5.5 KB | — | | .gitattributes | Repository | 1.6 KB | — | | merges.txt | Tokenizer | 1.7 MB | — | | tokenizer.json | Tokenizer | 11.4 MB | def76fb08697 | | tokenizer_config.json | Tokenizer | 5.4 KB | — | | vocab.json | Tokenizer | 2.8 MB | — | ## License and Download License apache-2.0 Access Open weights, no gate Download size 4.3 GB [Download from Qwen](https://modelscope.cn/models/Qwen/Qwen3-VL-Embedding-2B) Released by Qwen through ModelScope. [Read the license](https://www.apache.org/licenses/LICENSE-2.0). ## Built From - Derived from [Qwen/Qwen3-VL-2B-Instruct](https://savrn.com/models/qwen3-vl-2b-instruct) - Described by [arXiv:2601.04720](https://savrn.com/papers/qwen3-vl-embedding-and-qwen3-vl-reranker-a-unified-framework-for-state-of-the-art-multimod) ## Memory Requirements | Precision | Weights in memory | | --- | --- | | As published | 4.3 GB | | 16-bit | 4.3 GB | | 8-bit | 2.1 GB | | 4-bit | 1.1 GB | Weights only, from the published parameter count; the key-value cache and runtime add to this. ## Built on This Model - Quantized from[Qwen3-VL-Embedding-2B-Q8_0-GGUF](https://savrn.com/models/qwen3-vl-embedding-2b-q8-0-gguf) - Derived from[Qwen3-VL-Embedding-2B-Q8_0-GGUF](https://savrn.com/models/qwen3-vl-embedding-2b-q8-0-gguf) ## Questions About Qwen3-VL-Embedding-2B ### How much GPU memory does Qwen3-VL-Embedding-2B need? About 5.1 GB at 16-bit and 1.3 GB at 4-bit: the weights (2.1B parameters) plus a working margin. A long context needs more. ### What is the cheapest GPU to run Qwen3-VL-Embedding-2B on? At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists. ### Can I use Qwen3-VL-Embedding-2B commercially? Yes. Qwen3-VL-Embedding-2B is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors. ### What is Qwen3-VL-Embedding-2B's context length? 262,144 tokens, from the maximum position embeddings in its published configuration. ## Similar Models Model · Sentence similarity ### [gte-Qwen2-1.5B-instruct](https://savrn.com/models/gte-qwen2-1-5b-instruct) [Alibaba-NLP](https://savrn.com/model-publishers/alibaba-nlp) gte-Qwen2-1.5B-instruct is the latest model in the gte (General Text Embedding) model family. The model is built on Qwen2-1.5B LLM model and use the same training data and strategies as the gte-Qwen2-7B-instruct model. The model incorporates several key advancements: - Integration of bidirectional attention mechanisms, enriching its contextual understanding. - Instruction tuning, applied solely on the query side for streamlined efficiency - Comprehensive training across a vast, multilingual text corpus spanning diverse domains and scenarios. This training leverages both weakly supervised and supervised data, ensuring the model's applicability across numerous languages and a wide array of… Open weights apache-2.0 1.8B parameters 131,072 tokens sentence-transformers [View model](https://savrn.com/models/gte-qwen2-1-5b-instruct) Model · Sentence similarity ### [Nemotron-3-Embed-1B-BF16](https://savrn.com/models/nemotron-3-embed-1b-bf16) [NVIDIA](https://savrn.com/model-publishers/nvidia) NVIDIA Nemotron 3 Embed Nemotron-3-Embed-1B-BF16 is a versatile text embedding model trained by NVIDIA and optimized for retrieval and semantic similarity tasks. It provides strong multilingual and cross-lingual retrieval capabilities and is designed to serve as a foundational component in text-based Retrieval-Augmented Generation (RAG) systems. This model was evaluated across 34 languages: English, Arabic, Assamese, Bengali, Bulgarian, Chinese, Danish, Dutch, Finnish, French, German, Hindi, Hinglish, Indonesian, Italian, Japanese, Korean, Malay, Marathi, Nepali, Norwegian, Persian, Portuguese, Romanian, Russian, Spanish, Swahili, Swedish, Tamil, Telugu, Thai, Ukrainian, Urdu, Vietnamese.… Open weights other 1.1B parameters 262,144 tokens sentence-transformers [View model](https://savrn.com/models/nemotron-3-embed-1b-bf16) Model · Sentence similarity ### [AX-Nemotron-3-Embed-1B-CUDA-AXQ-NVFP4-W4A4](https://savrn.com/models/ax-nemotron-3-embed-1b-cuda-axq-nvfp4-w4a4) [AutomatosX](https://savrn.com/model-publishers/automatosx) AXQuant CUDA NVFP4 W4A4 mixed precision development preview. Converted from the original upstream BF16 source, without AWQ. Native E2M1 FP4 weights and inputs use per-16 E4M3FN scales and FP32 global scales. This is a retrieval embedding checkpoint, with no MTP or generative claim. Eligible attention and MLP matrices use NVFP4 W4A4. Embeddings and norms retain original BF16. Saved query: / passage: prefixes, bidirectional attention, mean pooling and L2 normalization. The native vLLM Ministral alias retains iscausal=false and every actual attention layer is verified as encoder-only. Factory conversion uses AXQuant's NumPy reference RTN encoder, independently checked against preserved BF16… Open weights other 1.1B parameters 262,144 tokens vllm [View model](https://savrn.com/models/ax-nemotron-3-embed-1b-cuda-axq-nvfp4-w4a4) Model · Sentence similarity ### [embeddingmagibu2-a100](https://savrn.com/models/embeddingmagibu2-a100) [Ali Bayram](https://savrn.com/model-publishers/alibayram) embeddingmagibu2 is a Turkish-focused multilingual embedding model built from google/embeddinggemma-2 and the Turkish-optimized magibu/Altus tokenizer. Offline distillation from google/embeddinggemma-2 is in progress; this table is updated automatically at every release. Latest release: phase5-train-new-done. Phases: test/new → test/full → validation/new → validation/full → train/new → train/full (new: only the new Turkish token embeddings train; full: all embeddings + text backbone at a low learning rate). Metrics on 2,048 held-out validation rows: cosine to the teacher's document embeddings for Turkish and for the other languages, their mean (score), and STSbTR test Spearman. It follows… Access requested at publisher apache-2.0 744M parameters sentence-transformers [View model](https://savrn.com/models/embeddingmagibu2-a100) Model · Sentence similarity ### [AX-Qwen3-Embedding-0.6B-CUDA-AXQ-NVFP4-W4A4](https://savrn.com/models/ax-qwen3-embedding-0-6b-cuda-axq-nvfp4-w4a4) [AutomatosX](https://savrn.com/model-publishers/automatosx) AXQuant CUDA NVFP4 W4A4 mixed precision development preview. Converted from the original upstream BF16 source, without AWQ. Native E2M1 FP4 weights and inputs use per-16 E4M3FN scales and FP32 global scales. This is a retrieval embedding checkpoint, with no MTP or generative claim. All attention projections and the first/last two MLP blocks retain original BF16. Remaining MLP matrices use NVFP4 W4A4. Saved query instruction, last-token pooling, exactly one (151643), and L2 normalization. The chat (151645) is not appended. Factory conversion uses AXQuant's NumPy reference RTN encoder, independently checked against preserved BF16 source tensors. Source and pooling assets, calibration and… Open weights apache-2.0 596M parameters 32,768 tokens vllm [View model](https://savrn.com/models/ax-qwen3-embedding-0-6b-cuda-axq-nvfp4-w4a4) Model · Sentence similarity ### [snowflake-arctic-embed-l-v2.0](https://savrn.com/models/snowflake-arctic-embed-l-v2-0) [Snowflake](https://savrn.com/model-publishers/snowflake) 12/11/2024: Release of Technical Report - 12/04/2024: Release of snowflake-arctic-embed-l-v2.0 and snowflake-arctic-embed-m-v2.0 our newest models with multilingual workloads in mind. Snowflake arctic-embed-l-v2.0 is the newest addition to the suite of embedding models Snowflake has released optimizing for retrieval performance and inference efficiency. Arctic Embed 2.0 introduces a new standard for multilingual embedding models, combining high-quality multilingual text retrieval without sacrificing performance in English. Released under the permissive Apache 2.0 license, Arctic Embed 2.0 is ideal for applications that demand reliable, enterprise-grade multilingual search and retrieval at… Open weights apache-2.0 568M parameters 8,194 tokens sentence-transformers [View model](https://savrn.com/models/snowflake-arctic-embed-l-v2-0) ## Qwen [All models and datasets](https://savrn.com/model-publishers/qwen) ## Versions - [9f2f7e710d6d](https://savrn.com/models/qwen3-vl-embedding-2b/versions/9f2f7e710d6d) · current 2026-10-14 ## Explore More - [All sentence similarity models](https://savrn.com/models/tasks/sentence-similarity) - [All models under apache-2.0](https://savrn.com/models/licenses/apache-2-0) - [Model comparisons](https://savrn.com/models/comparisons) - [The model directory](https://savrn.com/models) - [Open model prices by host](https://savrn.com/ai-index/pricing/open-models) ## Source - Repository metadata, read 2026-10-19. - [Qwen website](https://qwen.ai/) - [Hugging Face record](https://huggingface.co/Qwen/Qwen3-VL-Embedding-2B) - [How the hub is built](https://savrn.com/model-hub/methodology)