rank1-7b
jhu-clsp/rank1-7b
A popular open reranker model, with 300K downloads a month. gigarouter benchmarks and hosts it as an OpenAI-compatible API.
about this model
Overview
rank1-7b is a 7B parameter reasoning reranker trained from Qwen2.5-7B. Unlike traditional rerankers that output a relevance score directly, rank1 first generates an explicit reasoning chain within a thinking...response section, then produces a binary relevance judgment (true/false) and a confidence score derived from the logits of those tokens. This test-time compute approach helps the model break down complex relevance decisions into logical steps.
Key Strengths
- Generates interpretable reasoning chains before making relevance judgments.
- Returns both a binary label and a calibrated confidence score.
- Effective for nuanced topics where direct scoring may be insufficient.
Best Use Cases
Reranking document lists in information retrieval pipelines, especially for queries requiring reasoning about relevance rather than simple keyword matching.
Performance
The model demonstrates strong performance on retrieval benchmarks, particularly on tasks requiring complex reasoning. Detailed benchmark results and comparisons are available in the paper and the official repository.
Model Variants
| Model | Base | Description |
|---|---|---|
| rank1-0.5b | Qwen2.5-0.5B | Smallest variant (0.5B parameters) |
| rank1-1.5b | Qwen2.5-1.5B | Smaller variant (1.5B parameters) |
| rank1-3b | Qwen2.5-3B | Smaller variant (3B parameters) |
| rank1-7b | Qwen2.5-7B | Current model (7B parameters) |
| rank1-14b | Qwen2.5-14B | Larger variant (14B parameters) |
| rank1-32b | Qwen2.5-32B | Largest variant (32B parameters) |
| rank1-mistral-2501-24b | Mistral-Small 2501 24B | Trained from Mistral base |
| rank1-llama3-8b | Llama 3.1 8B | Trained from Llama 3.1 |
Quantized variants (AWQ) are also available for several of the above models. See the model card for details.
We're benchmarking and onboarding rank1-7b as a hosted, OpenAI-compatible API. Sign in for free credit and be ready when it lands, or tell us you want it and we'll prioritize it.