---
title: google/siglip-so400m-patch14-384
description: SigLIP model pre-trained on WebLi at resolution 384x384. It was introduced in the paper Sigmoid Loss for Language Image Pre-Training by Zhai. SigLIP, 878M parameters.
canonical_url: https://superlinked.com/models/google-siglip-so400m-patch14-384
last_updated: 2026-05-25
---

# google/siglip-so400m-patch14-384

SigLIP model pre-trained on WebLi at resolution 384x384. It was introduced in the paper Sigmoid Loss for Language Image Pre-Training by Zhai et al. and first released in this repository.

Source: [google/siglip-so400m-patch14-384 on HuggingFace](https://huggingface.co/google/siglip-so400m-patch14-384)

## Overview

| Field | Value |
|-------|-------|
| Architecture | SigLIP |
| Parameters | 878M |
| Tasks | Encode |
| Outputs | Dense |
| Dimensions | Dense: 1,152 |
| Max sequence length | 64 tokens |
| License | apache-2.0 |
| Inputs | text, image |

## Benchmarks

### Flickr30kI2TRetrieval

Domain: general · Task: retrieval · Language: en

Image-to-text retrieval: retrieve captions from images

Corpus: 31,783 · Queries: 1,000

**Quality:** ndcg at 10: 0.9001 · map at 10: 0.8364 · mrr at 10: 0.9663

**Performance (L4-SPOT b1 c8):** Corpus 202 tok/s · Corpus p50 523.6ms · Query 9.7 img/s · Query p50 711.3ms

**Performance (L4 b1 c16):** Corpus 597 tok/s · Corpus p50 381.3ms · Query 5.7 mpix/s · Query p50 459.0ms

[Reference](https://shannon.cs.illinois.edu/DenotationGraph/)