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

← Catalog

NovaSearch/stella_en_400M_v5

Open comparison →

Primitive: /encode · Encode · ModernBERT

We released a Jasper and Stella model technology report and code.(2025.1)

Dense

Overview

Hardware: — drives latency, throughput & cost

Size435M params
Tasks /encode
Licensemit
Latency116 ms
Throughput27.1K tok/s
Cost$0.0082 /1M tok

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

Embedding

Output typesDense
Dimensionsdense: 1,024
Max sequence length512
Inputstext

Benchmarks

CQADupstackPhysicsRetrieval

scientific retrieval en

Duplicate question retrieval from StackExchange Physics

Corpus: 38,314 Queries: 1,039
Quality
ndcg at 10 0.4836
map at 10 0.4156
mrr at 10 0.4717
Performance L4 b1 c16
Corpus 24.3K tok/s
Corpus p50 84.2ms
Query 2.5K tok/s
Query p50 56.9ms
Reference →

CosQA

technology retrieval en

Code search with natural language queries

Corpus: 6,267 Queries: 500
Quality
ndcg at 10 0.3216
map at 10 0.2529
mrr at 10 0.2719
Performance L4 b1 c16
Corpus 12.8K tok/s
Corpus p50 67.0ms
Query 1.4K tok/s
Query p50 59.0ms
Reference →

FiQA2018

finance retrieval en

Financial opinion mining and question answering

Corpus: 57,599 Queries: 648
Quality
ndcg at 10 0.5589
map at 10 0.4742
mrr at 10 0.6391
Performance L4 b1 c16
Corpus 24.7K tok/s
Corpus p50 97.7ms
Query 2.6K tok/s
Query p50 59.9ms
Reference →

LegalBenchConsumerContractsQA

legal retrieval en

Question answering on consumer contracts

Corpus: 153 Queries: 396
Quality
ndcg at 10 0.8107
map at 10 0.7633
mrr at 10 0.7645
Performance L4 b1 c16
Corpus 39.7K tok/s
Corpus p50 209.7ms
Query 3.6K tok/s
Query p50 61.0ms
Reference →

NFCorpus

medical retrieval en

Biomedical literature search from NutritionFacts.org

Corpus: 3,593 Queries: 323
Quality
ndcg at 10 0.4125
map at 10 0.1615
mrr at 10 0.6096
Performance L4 b1 c16
Corpus 30.3K tok/s
Corpus p50 164.6ms
Query 1.1K tok/s
Query p50 68.1ms
Reference →

NanoFiQA2018Retrieval

finance retrieval en

Smaller subset of the FiQA financial QA dataset

Quality
ndcg at 10 0.6158
map at 10 0.5412
mrr at 10 0.6537
Performance L4 b1 c16
Corpus 25.2K tok/s
Corpus p50 103.9ms
Query 2.7K tok/s
Query p50 58.3ms
Reference →

SCIDOCS

scientific retrieval en

Citation prediction, document classification, and recommendation for scientific papers

Corpus: 25,656 Queries: 1,000
Quality
ndcg at 10 0.2501
map at 10 0.1523
mrr at 10 0.4130
Performance L4 b1 c16
Corpus 27.1K tok/s
Corpus p50 115.7ms
Query 2.6K tok/s
Query p50 57.4ms
Reference →

SciFact

scientific retrieval en

Scientific claim verification using research literature

Corpus: 5,183 Queries: 300
Quality
ndcg at 10 0.7842
map at 10 0.7358
mrr at 10 0.7455
Performance L4 b1 c16
Corpus 28.4K tok/s
Corpus p50 156.1ms
Query 3.4K tok/s
Query p50 60.0ms
Reference →

StackOverflowQA

technology retrieval en

Programming question answering from Stack Overflow

Corpus: 19,931 Queries: 1,994
Quality
ndcg at 10 0.8666
map at 10 0.8439
mrr at 10 0.8439
Performance L4 b1 c16
Corpus 28.2K tok/s
Corpus p50 134.6ms
Query 31.9K tok/s
Query p50 153.3ms
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 2.1K

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.