Why did we open-source our inference engine? Read the post

← Catalog

Qwen/Qwen3-VL-Reranker-2B

Open comparison →

Primitive: /score · Score · Qwen3-VL

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.

MultimodalLong context

Overview

Hardware: — drives latency, throughput & cost

Size2.1B params
Tasks /score
Licenseapache-2.0
Latency173 ms
Throughput258 tok/s
Cost$0.861 /1M tok

Cost is approximate — computed from list GPU prices; your actual price depends on the provider you deploy SIE with.

Scoring

Inputstext · image
Max sequence length32,768

Benchmarks

AskUbuntuDupQuestions

technology reranking en

Duplicate question detection from AskUbuntu

Corpus: 6,743 Queries: 360
Quality
ndcg at 10 0.6657
map at 10 0.5094
mrr at 10 0.7811
Performance L4 b1 c4
Corpus 267 tok/s
Corpus p50 173.4ms
Query 258 tok/s
Query p50 173.4ms
Performance RTX-PRO-6000-Blackwell-Server-Edition b1 c16
Corpus 1.7K tok/s
Corpus p50 61.2ms
Query 1.6K tok/s
Query p50 61.2ms
Reference →

CQADupstackPhysicsRetrieval

scientific retrieval en

Duplicate question retrieval from StackExchange Physics

Corpus: 38,314 Queries: 1,039
Quality
map at 10 0.4749
mrr at 10 0.5412
ndcg at 10 0.5389
Reference →

CosQA

technology retrieval en

Code search with natural language queries

Corpus: 6,267 Queries: 500
Quality
map at 10 0.2650
mrr at 10 0.2793
ndcg at 10 0.3412
Reference →

FiQA2018

finance retrieval en

Financial opinion mining and question answering

Corpus: 57,599 Queries: 648
Quality
map at 10 0.4210
mrr at 10 0.5909
ndcg at 10 0.5046
Reference →

LegalBenchConsumerContractsQA

legal retrieval en

Question answering on consumer contracts

Corpus: 153 Queries: 396
Quality
map at 10 0.8473
mrr at 10 0.8478
ndcg at 10 0.8781
Reference →

NFCorpus

medical retrieval en

Biomedical literature search from NutritionFacts.org

Corpus: 3,593 Queries: 323
Quality
map at 10 0.3121
mrr at 10 0.6392
ndcg at 10 0.4173
Reference →

SCIDOCS

scientific retrieval en

Citation prediction, document classification, and recommendation for scientific papers

Corpus: 25,656 Queries: 1,000
Quality
map at 10 0.1407
mrr at 10 0.3927
ndcg at 10 0.2298
Reference →

SciFact

scientific retrieval en

Scientific claim verification using research literature

Corpus: 5,183 Queries: 300
Quality
map at 10 0.7498
mrr at 10 0.7637
ndcg at 10 0.7870
Reference →

StackOverflowQA

technology retrieval en

Programming question answering from Stack Overflow

Corpus: 19,931 Queries: 1,994
Quality
map at 10 0.9449
mrr at 10 0.9451
ndcg at 10 0.9506
Reference →

Vidore3FinanceEnRetrieval

finance retrieval en

Visual document retrieval on financial reports

Quality
ndcg at 10 0.6937
map at 10 0.5974
mrr at 10 0.8018
Reference →

Vidore3FinanceEnRetrieval (lang eng, candidates: Qwen3-VL-Embedding-2B, k=50)

finance retrieval en

Visual document retrieval on financial reports

Performance RTX-PRO-6000-Blackwell-Server-Edition b1 c16
Corpus 40.7 mpix/s
Corpus p50 684.8ms
Query 230 tok/s
Query p50 684.8ms
Reference →

Open source inference for agents

Open-source inference for the models behind your agents. Run it yourself, or let us run it for you.

Github 3.3K

Contact us

Tell us about your use case and we'll get back to you shortly.

Apply for an inference grant

Free capacity on our hosted cluster for selected projects. Tell us what you run and we reply by email.