Google says Gemini 4 Argon ranks well behind Claude and GPT-6 for Android coding

Description

Google has updated its Android Bench list for showing how good AI models are at Android coding assistance, with Gemini 4 Argon ranking way behind that of the latest Claude and GPT models.

The Android Bench was introduced early this year as a way for Android developers to get a quick way to see what AI models were best suited to Android coding assistance based on a test from Google. That test took into account aspects of AI use that are specific to Android coding and best practices in Android development. As Google has continued its tests and ranking new models, it has also included the average cost as a part of the benchmark so that developers can make the best decision for their needs.

With the October 26 update, Google has added the reasoning level used for each model, as well as adding results for OpenAI’s GPT-6 Sol, GPT-6.1 Sol, Claude Sonnet 5, and Claude Sonnet 5.5, as well as Google’s own Gemini 4 Argon.

Currently topping the chart is Claude Sonnet 5.5, with a pass rate of 40% and a completion rate of 80.9%. Claude Opus 5.5, added to the list in early October, comes in second place.

OpenAI’s GPT-6.1 Sol might be the most interesting addition here, coming in at a pass rate of 29.3% and completion rate of 82.8%, but a cost of just $99.50 compared to Claude’s $1,574.90 cost for its first place finish.

But it’s Google’s own Gemini 4 Argon that makes a rather underwhelming debut, sitting well behind other frontier models in fifth place. It lags behind Claude Fable 5.1 at a mere 21.3% pass rate, 81.8% completion rate, and an average cost of $267.70.

In a report shortly after Google’s announcement of Gemini 4 Argon, some internal voices claimed that the model “struggles to handle certain coding tasks,” though Google stood by its capabilities.