FlashRank Pro
BaseApache 2.0149M ModernBERT Cross-Encoder Reranker for Enterprise RAG
FlashRank Pro Base is a 149m modernbert cross-encoder reranker for enterprise rag. Operates across Parameters: 149M · Base: ModernBERT-base · Latency: <50ms on CPU · Context: 8,192 tokens · Architecture: Cross-Encoder · Hardware: CPU / Edge · License: Apache 2.0. License: Apache 2.0.
A 149M-parameter cross-encoder reranker fine-tuned on ModernBERT-base for high-precision retrieval-augmented generation. Evaluates query-document relevance in under 50ms on CPU with an 8K context window. Engineered to separate hard negatives and eliminate hallucinations in legal and financial RAG pipelines without requiring a dedicated GPU.
- ✓Sub-50ms CPU inference for real-time search and retrieval
- ✓ModernBERT backbone with full 8,192 token context window
- ✓Fine-tuned on MS MARCO hard negatives for precision reranking
- ✓Powers zero-GPU on-premise RAG in LexRAG and Evolucent AI
Get started in minutes
pip install transformers torchfrom transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch
tokenizer = AutoTokenizer.from_pretrained("eulogik/flashrank-pro-base")
model = AutoModelForSequenceClassification.from_pretrained("eulogik/flashrank-pro-base")
query = "What are the penalties under DPDP Section 16?"
doc = "Section 16 outlines penalties up to 250 crore for significant data fiduciaries..."
inputs = tokenizer(query, doc, return_tensors="pt", truncation=True, max_length=512)
with torch.no_grad():
scores = model(**inputs).logits
print("Relevance Score:", scores.item())Cite this Model & Architecture
If you utilize FlashRank Pro in your academic research, benchmarks, or enterprise deployments, please cite our open-weight publication:
@misc{eulogik2026flashrankpro,
title = {FlashRank Pro Base: 149M ModernBERT Cross-Encoder Reranker for Enterprise RAG},
author = {Kishore, Gautam and Eulogik Systems Engineering},
year = {2026},
publisher = {Eulogik},
howpublished = {\url{https://eulogik.com/models/flashrank-pro}}
}Need help deploying FlashRank Pro?
We offer consulting, custom fine-tuning, and on-premise deployment for all our models.