[GH-ISSUE #4106] 配置OPENAI_API_BASE_URL之后，前端无法访问 #13501

New Issue

GiteaMirror · 2026-04-19T20:13:27-05:00

GiteaMirror commented

2026-04-19 20:13:27 -05:00

Originally created by @shouzheng on GitHub (Jul 25, 2024).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/4106

Bug Report

Description

Bug Summary:

在本地服务器部署了OpenAI API兼容的大模型服务（使用vllm），然后使用下面的命令启动open-webui，在注册、登录之后，前端无法正常工作，页面空白。

启动命令：

docker run -d -p 3030:8080 \
-e OPENAI_API_KEY=NULL \
-e OPENAI_API_BASE_URL=http://10.3.219.78:49423/v1 \
-v open-webui:/app/backend/data \
--name open-webui \
--restart always ghcr.io/open-webui/open-webui:main

报错内容：

Models.svelte:25 Uncaught (in promise) TypeError: I.at is not a function
    at Models.svelte:25
    at Array.filter (<anonymous>)
    at Object.Bo.s.$$.update (Models.svelte:24)
    at Ot (Component.js:144)
    at new Lo (Models.svelte:30)
    at Si (MessageInput.svelte:339)
    at Ot (Component.js:148)
    at new Ti (MessageInput.svelte:90)
    at bn (Chat.svelte:1614)
    at Ua (Chat.svelte:1510)

Steps to Reproduce:
[Outline the steps to reproduce the bug. Be as detailed as possible.]

在服务器使用vllm启动大模型服务，使用命令如：

nohup python -m vllm.entrypoints.openai.api_server --served-model-name Qwen2-7B-Instruct --port 7777 --model ~/Qwen2-7B-Instruct > ~/logs/Qwen2-7B-Instruct.log 2>&1 &

使用下面的命令启动open-webui：

docker run -d -p 3030:8080 \
-e OPENAI_API_KEY=NULL \
-e OPENAI_API_BASE_URL=http://10.3.219.78:49423/v1 \
-v open-webui:/app/backend/data \
--name open-webui \
--restart always ghcr.io/open-webui/open-webui:main

该命令参考了Installation for OpenAI API Usage Only。

访问：localhost:3030页面，进行注册、登录
登录之后发现页面空白，打开浏览器的Console，发现有报错，报错内容：

Models.svelte:25 Uncaught (in promise) TypeError: I.at is not a function
    at Models.svelte:25
    at Array.filter (<anonymous>)
    at Object.Bo.s.$$.update (Models.svelte:24)
    at Ot (Component.js:144)
    at new Lo (Models.svelte:30)
    at Si (MessageInput.svelte:339)
    at Ot (Component.js:148)
    at new Ti (MessageInput.svelte:90)
    at bn (Chat.svelte:1614)
    at Ua (Chat.svelte:1510)

根据调试，关键代码在这里：

export let prompt = '';
# ......
$: filteredModels = $models
    .filter((p) =>
        p.name.toLowerCase().includes(prompt.toLowerCase().split(' ')?.at(0)?.substring(1) ?? '')
    )
    .sort((a, b) => a.name.localeCompare(b.name));

在异常时，p.name='Qwen2-7B-Instruct', prompt=''。

Expected Behavior:
[Describe what you expected to happen.]

前端页面可以正常访问。

Originally created by @shouzheng on GitHub (Jul 25, 2024). Original GitHub issue: https://github.com/open-webui/open-webui/issues/4106 # Bug Report ## Description **Bug Summary:** 在本地服务器部署了OpenAI API兼容的大模型服务（使用vllm），然后使用下面的命令启动open-webui，在注册、登录之后，前端无法正常工作，页面空白。启动命令： ```sh docker run -d -p 3030:8080 \ -e OPENAI_API_KEY=NULL \ -e OPENAI_API_BASE_URL=http://10.3.219.78:49423/v1 \ -v open-webui:/app/backend/data \ --name open-webui \ --restart always ghcr.io/open-webui/open-webui:main ``` 报错内容： ``` Models.svelte:25 Uncaught (in promise) TypeError: I.at is not a function at Models.svelte:25 at Array.filter (<anonymous>) at Object.Bo.s.$$.update (Models.svelte:24) at Ot (Component.js:144) at new Lo (Models.svelte:30) at Si (MessageInput.svelte:339) at Ot (Component.js:148) at new Ti (MessageInput.svelte:90) at bn (Chat.svelte:1614) at Ua (Chat.svelte:1510) ``` **Steps to Reproduce:** [Outline the steps to reproduce the bug. Be as detailed as possible.] 1. 在服务器使用vllm启动大模型服务，使用命令如： ```sh nohup python -m vllm.entrypoints.openai.api_server --served-model-name Qwen2-7B-Instruct --port 7777 --model ~/Qwen2-7B-Instruct > ~/logs/Qwen2-7B-Instruct.log 2>&1 & ``` 2. 使用下面的命令启动open-webui： ```sh docker run -d -p 3030:8080 \ -e OPENAI_API_KEY=NULL \ -e OPENAI_API_BASE_URL=http://10.3.219.78:49423/v1 \ -v open-webui:/app/backend/data \ --name open-webui \ --restart always ghcr.io/open-webui/open-webui:main ``` 该命令参考了[Installation for OpenAI API Usage Only](https://docs.openwebui.com/getting-started/#installation-for-openai-api-usage-only)。 3. 访问：localhost:3030页面，进行注册、登录 4. 登录之后发现页面空白，打开浏览器的Console，发现有报错，报错内容： ``` Models.svelte:25 Uncaught (in promise) TypeError: I.at is not a function at Models.svelte:25 at Array.filter (<anonymous>) at Object.Bo.s.$$.update (Models.svelte:24) at Ot (Component.js:144) at new Lo (Models.svelte:30) at Si (MessageInput.svelte:339) at Ot (Component.js:148) at new Ti (MessageInput.svelte:90) at bn (Chat.svelte:1614) at Ua (Chat.svelte:1510) ``` 根据调试，关键代码在这里： ``` export let prompt = ''; # ...... $: filteredModels = $models .filter((p) => p.name.toLowerCase().includes(prompt.toLowerCase().split(' ')?.at(0)?.substring(1) ?? '') ) .sort((a, b) => a.name.localeCompare(b.name)); ``` 在异常时，p.name='Qwen2-7B-Instruct', prompt=''。 **Expected Behavior:** [Describe what you expected to happen.] 前端页面可以正常访问。

GiteaMirror closed this issue

2026-04-19 20:13:28 -05:00

GiteaMirror referenced this issue

2026-04-19 22:44:31 -05:00

[GH-ISSUE #13501] feat: What are your thoughts on removing requirements.txt and using only pyproject.toml? #16930

GiteaMirror referenced this issue

2026-04-25 06:22:55 -05:00

[GH-ISSUE #13501] feat: What are your thoughts on removing requirements.txt and using only pyproject.toml? #32459

GiteaMirror referenced this issue

2026-05-05 17:41:08 -05:00

[GH-ISSUE #13501] feat: What are your thoughts on removing requirements.txt and using only pyproject.toml? #55596

GiteaMirror referenced this issue