Simon Willison’s Weblog

Google Gemini URL Context (via) New feature in the Gemini API: you can now enable a url_context tool which the models can use to request the contents of URLs as part of replying to a prompt.

I released llm-gemini 0.25 with a new -o url_context 1 option adding support for this feature. You can try it out like this:

llm install -U llm-gemini
llm keys set gemini # If you need to set an API key
llm -m gemini-2.5-flash -o url_context 1 \
  'Latest headline on simonwillison.net'

Tokens from the fetched content are charged as input tokens. Use llm logs -c --usage to see that token count:

# 2025-08-18T23:52:46    conversation: 01k2zsk86pyp8p5v7py38pg3ge id: 01k2zsk17k1d03veax49532zs2

Model: **gemini/gemini-2.5-flash**

## Prompt

Latest headline on simonwillison.net

## Response

The latest headline on simonwillison.net as of August 17, 2025, is "TIL: Running a gpt-oss eval suite against LM Studio on a Mac.".

## Token usage

9,613 input, 87 output, {"candidatesTokenCount": 57, "promptTokensDetails": [{"modality": "TEXT", "tokenCount": 10}], "toolUsePromptTokenCount": 9603, "toolUsePromptTokensDetails": [{"modality": "TEXT", "tokenCount": 9603}], "thoughtsTokenCount": 30}

I intercepted a request from it using django-http-debug and saw the following request headers:

Accept: */*
User-Agent: Google
Accept-Encoding: gzip, br

The request came from 192.178.9.35, a Google IP. It did not appear to execute JavaScript on the page, instead feeding the original raw HTML to the model.

# 11:59 pm / google, ai, generative-ai, llms, llm, gemini