GPTQ Quantized Qwen/Qwen3-Reranker-4B with Ultrachat, THUIR/T2Ranking and m-a-p/COIG-CQIA for calibration set. VRAM Usage: 17430M -> 11000M (w/o FA2, according to Embedding model's result). I think <5% accuracy, further evaluation on the way... The Embedding one shows ~0.7%. pip install compressed-tensors optimum and auto-gptq / gptqmodel, then goto the official usage guide.
pipelinetag: text-classification - sentence-transformers - vidore - reranker - qwen2vl - multilingual basemodel: libraryname: transformers jina-reranker-m0 is our new multilingual multimodal reranker model for ranking visual documents across multiple…
Parameters2.4B
Context32,768
Weights4.9 GB
Licensecc-by-nc-4.0
AccessOpen weights
Monthly Downloads394.1k
Runs On
What it takes to serve jina-reranker-m0 (2.4B 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 | 4.9 GB | 5.9 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 2.4 GB | 2.9 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 1.2 GB | 1.5 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x 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, read Sep 18, 2026.
SAVRN's Notes on jina-reranker-m0
Put this one in front of a document search, not behind a chat window. jina-reranker-m0 takes a query plus a stack of page images, tables, infographics and mixed layouts in over 29 languages and returns them ordered by relevance. At 2.4 billion parameters it needs 5.9 GB at 16-bit and 1.5 GB at 4-bit, so the cheapest priced slot, one MI300X with 192 GB at $1.85 an hour, is best shared with a larger model that already holds the card.
The license is the stopper. CC BY-NC 4.0 allows sharing and adapting with credit for non-commercial purposes only; a ranking service for paying customers needs separate permission from the rights holder, and that call comes before procurement. Also check the 32,768-token context, which caps how many pages score in one pass, and the derivation from Qwen2-VL-2B-Instruct, whose own terms a buyer should read alongside Jina's.
Model Card
pipelinetag: text-classification - sentence-transformers - vidore - reranker - qwen2vl - multilingual basemodel: libraryname: transformers jina-reranker-m0 is our new multilingual multimodal reranker model for ranking visual documents across multiple languages: it accepts a query alongside a collection of visually rich document images, including pages with text, figures, tables, infographics, and various layouts across multiple domains and over 29 languages. It outputs a ranked list of documents ordered by their relevance to the input query. Compared to jina-reranker-v2-base-multilingual, jina-reranker-m0 also improves text reranking for multilingual content, long documents, and code…
Excerpt from the card by Jina AI, licensed cc-by-nc-4.0.
Configuration
- Architecture
- JinaVLForRanking
- Context length (tokens)
- 32,768
- Layers
- 28
- Hidden size
- 1,536
- Feed-forward size
- 8,960
- Attention heads
- 12
- Key/value heads
- 2
- Vocabulary size
- 151,936
- Sliding window (tokens)
- 32,768
- RoPE base
- 1e+06
- Stored precision
- bfloat16
- Model type
- qwen2_vl
Identity and Version
- Repository
- jinaai/jina-reranker-m0
- Publisher
- Jina AI
- Task
- Text classification
- Modality
- Text
- Library
- transformers
- Parameters
- 2.4B parameters
- Languages
- Not stated by the source
- Revision
- 94bfe0aeb2d4dd7978362699cddd5893d4e0adc8
- First published
- 2025-03-27
- Last updated
- 2026-04-09
Files and Weights
18 files, 4.9 GB in total. The weights are 1 file totalling 4.9 GB in safetensors.
Weights1 file · 4.9 GB
Configuration10 files · 15.3 KB
Tokenizer4 files · 15.9 MB
Documentation1 file · 24.3 KB
Other1 file · 808 B
Repository1 file · 1.6 KB
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| model.safetensors | Weights | 4.9 GB | 1d0a7b5fd096 |
| added_tokens.json | Configuration | 392 B | — |
| config.json | Configuration | 1.2 KB | — |
| config_sentence_transformers.json | Configuration | 229 B | — |
| custom_transformer.py | Configuration | 1.6 KB | — |
| generation_config.json | Configuration | 261 B | — |
| modeling.py | Configuration | 10.1 KB | — |
| modules.json | Configuration | 115 B | — |
| preprocessor_config.json | Configuration | 314 B | — |
| sentence_bert_config.json | Configuration | 524 B | — |
| special_tokens_map.json | Configuration | 613 B | — |
| README.md | Documentation | 24.3 KB | — |
| chat_template.jinja | Other | 808 B | — |
| .gitattributes | Repository | 1.6 KB | — |
| merges.txt | Tokenizer | 1.7 MB | — |
| tokenizer.json | Tokenizer | 11.4 MB | 091aa7594dc2 |
| tokenizer_config.json | Tokenizer | 3.3 KB | — |
| vocab.json | Tokenizer | 2.8 MB | — |
License and Download
- License
- cc-by-nc-4.0
- Access
- Open weights, no gate
- Download size
- 4.9 GB
Download from Jina AI
Released by Jina AI through its official repository on Hugging Face. Read the license.
Built From
- Derived from Qwen/Qwen2-VL-2B-Instruct
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 4.9 GB |
| 16-bit | 4.9 GB |
| 8-bit | 2.4 GB |
| 4-bit | 1.2 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About jina-reranker-m0
How much GPU memory does jina-reranker-m0 need?
About 5.9 GB at 16-bit and 1.5 GB at 4-bit: the weights (2.4B parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run jina-reranker-m0 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 jina-reranker-m0 commercially?
Not without separate permission. jina-reranker-m0 is released under Creative Commons Attribution-NonCommercial 4.0. CC BY-NC 4.0 permits sharing and adapting with credit for non-commercial purposes only. Commercial use needs separate permission from the rights holder.
What is jina-reranker-m0's context length?
32,768 tokens, from the maximum position embeddings in its published configuration.
Similar Models
More details please refer to our Github: FlagEmbedding. Different from embedding model, reranker uses question and document as input and directly output similarity instead of embedding. You can get a relevance score by inputting query and passage to the reranker. And the score can be mapped to a float value in [0,1] by sigmoid function. You can select the model according your senario and resource. - For multilingual, utilize BAAI/bge-reranker-v2-m3 and BAAI/bge-reranker-v2-gemma - For Chinese or English, utilize BAAI/bge-reranker-v2-m3 and BAAI/bge-reranker-v2-minicpm-layerwise. - For efficiency, utilize BAAI/bge-reranker-v2-m3 and the low layer of BAAI/bge-reranker-v2-minicpm-layerwise.…
Opir-multitask-large is the English, highest-accuracy multi-task checkpoint in the Opir family: an encoder-based GLiClass guardrail model for real-time LLM safety filtering. It supports binary safe/unsafe classification, toxicity detection, jailbreak and prompt-injection detection, and zero-shot harmful-content categorization over a hierarchical safety taxonomy. This card is for knowledgator/opir-multitask-large. The model is used through GLiClass zero-shot classification: pass text plus the candidate labels you want scored. Use single-label mode for binary safe/unsafe decisions and multi-label mode for taxonomy, toxicity, jailbreak, or custom policy labels. Use multi-label mode when you…
roberta-large-mnli is the RoBERTa large model fine-tuned on the Multi-Genre Natural Language Inference (MNLI) corpus. The model is a pretrained model on English language text using a masked language modeling (MLM) objective. Use the code below to get started with the model. The model can be loaded with the zero-shot-classification pipeline like so: You can then use this pipeline to classify sequences into any of the class names you specify. For example: This fine-tuned model can be used for zero-shot classification tasks, including zero-shot sentence-pair classification (see the GitHub repo for examples) and zero-shot sequence classification. The model should not be used to intentionally…
LLM-powered applications are susceptible to prompt attacks, which are prompts intentionally designed to subvert the developer’s intended behavior of the LLM. Categories of prompt attacks include prompt injection and jailbreaking: - Prompt Injections are inputs that exploit the concatenation of untrusted data from third parties and users into the context window of a model to get a model to execute unintended instructions. - Jailbreaks are malicious instructions designed to override the safety and security features built into a model. Prompt Guard is a classifier model trained on a large corpus of attacks, capable of detecting both explicitly malicious prompts as well as data that contains…
This model is a fine-tuned version of xlm-roberta-base on the Language Identification dataset. This model is an XLM-RoBERTa transformer model with a classification head on top (i.e. a linear layer on top of the pooled output). For additional information please refer to the xlm-roberta-base model card or to the paper Unsupervised Cross-lingual Representation Learning at Scale by Conneau et al. You can directly use this model as a language detector, i.e. for sequence classification tasks. Currently, it supports the following 20 languages: arabic (ar), bulgarian (bg), german (de), modern greek (el), english (en), spanish (es), french (fr), hindi (hi), italian (it), japanese (ja), dutch (nl)…