VLLM and openai api capability support

[devops724]

Problem Description

Hi is there any support to openai api capability support provide by vllm i want test some models with browser use like qwen-vl model the only way i found os inference with vlm models vllm serve and connect browser-use to open it currenly after few step i get error like this Attempted to assign 1794 = 1794 multimodal tokens to 0 placeholders and vllm crash best regards

Proposed Solution

add browser-use support openapi capability support provide by vllm

Alternative Solutions

No response

Additional Context

No response

[ChirawatCh]

I am also waiting it to support vLLM API endpoint.

[seasoncool]

mark here +1

[dcoracle]

+1

[timonrieger]

+1

[love01211]

+1

[love01211]

@devops724 You can set up chatopenai and then configure the relevant information of vllm. It can be used in actual tests

import asyncio
import os
import sys

sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))

from dotenv import load_dotenv

load_dotenv()

from langchain_openai import ChatOpenAI

from browser_use import Agent

# Initialize the model
llm = ChatOpenAI(
        # ADD
        base_url="http://192.168.114.114:18080/v1",
        # Change
	model='Qwen3-4B',
	temperature=0.0,
)
task = 'Go to kayak.com and find the cheapest one-way flight from Zurich to San Francisco in 3 weeks.'
# Set it to True or false depending on whether it is a multimodal model
agent = Agent(task=task, llm=llm, use_vision=False)


async def main():
	await agent.run()


if __name__ == '__main__':
	asyncio.run(main())