# Vultron Retriever Core 4.5B

Vultr's reranker balancing accuracy and throughput for retrieval and RAG, with a 262K-token context. Input longer than the context length is rejected with a 400 (context_length_exceeded).

Model id: `vultron-retriever-core-qwen3.5-4.5b`
Context: 256K tokens
Quantization: bf16
Regions: ATL
Hugging Face: [vultr/VultronRetrieverCore-Qwen3.5-4.5B](https://huggingface.co/vultr/VultronRetrieverCore-Qwen3.5-4.5B)

## Pricing

- Tokens: $0.10 / 1M tokens

## Inputs

- text: context 256K tokens

## Output: rerank

| Parameter | Type | Values | Default |
| --- | --- | --- | --- |
| `top_n` | integer | 1–100 |  |
| `return_documents` | boolean |  |  |

## Endpoints

- [`POST /v1/rerank`](/docs/reference/create-rerank.md): Rerank Documents