> ## Content Index
> Fetch the complete content index at: https://www.testingcatalog.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# Google tests Gemini 3 Flash model on LM Arena
- URL: https://www.testingcatalog.com/google-tests-new-gemini-3-models-on-lm-arena/
- Published: 2025-12-11T15:21:54.000Z
- Updated: 2025-12-11T15:30:36.000Z
- Description: Google is testing Fierce Falcon and Ghost Falcon models on LM Arena, signalling a potential Gemini 3 Flash or Gemini 3 Pro GA launch.
- Author: Alexey Shabanov
- Tags: Gemini News, AI Rumours, Latest AI News

Google is actively testing two new models on LM Arena named **Fierce Falcon** and **Ghost Falcon**, which could represent the next stage in their [Gemini 3](https://www.testingcatalog.com/google-launches-gemini-3-models-across-its-platforms/) lineup. This phase fits Google's established pattern of pushing preview versions, followed by a higher-performance checkpoint that later becomes the general availability model. Historically, this approach allowed Google to quickly address early issues and offer notable improvements for wide release. These current moves seem timed to respond directly to OpenAI's [upcoming state-of-the-art models](https://www.testingcatalog.com/openai-debuts-robin-high-model-on-lm-arena-for-public-testing/) and their own [image generation upgrades](https://www.testingcatalog.com/openai-testing-new-image-2-models-on-lm-arena/), which have been quietly surfacing on LM Arena as well.

![Arrange the six numbers 2, 0, 1, 9, 20, and 19 in any order to form an 8-digit number (the first digit cannot be 0). How many different 8-digit numbers can be formed?](https://storage.ghost.io/c/2a/1b/2a1b1782-8506-4d7d-bf53-ad3fb52e2a0f/content/images/2025/12/LMArena-12-11-2025_02_32_PM.jpg)

LM Arena

With only days left in December, the competitive landscape is tightening. The presence of these unreleased Google models suggests that a public launch of an upgraded Gemini 3 Flash or Gemini 3 Pro GA is imminent, positioning it as a likely candidate for broader deployment just as OpenAI prepares its next major release. 

> Grok 4.2 is planned to arrive in 3 or 4 weeks. This release needs to be special.   
>  
> Soon 👀 [https://t.co/dwYh8MXyvJ](https://t.co/dwYh8MXyvJ?ref=testingcatalog.com) [pic.twitter.com/pmzNUd87xQ](https://t.co/pmzNUd87xQ?ref=testingcatalog.com)
> 
> — TestingCatalog News 🗞 (@testingcatalog) [December 7, 2025](https://twitter.com/testingcatalog/status/1997564220817408148?ref%5Fsrc=twsrc%5Etfw&ref=testingcatalog.com)

Meanwhile, Grok 4.2 from xAI is also anticipated in the next three weeks, although the specific timing remains unclear and may slip into the start of the new year. For researchers and early adopters, these developments mean the final ranking on LM Arena for 2025 is still uncertain, keeping the attention of the AI community and prediction markets like [Polymarket](https://polymarket.com/event/which-company-has-best-ai-model-end-of-2025?tid=1765466264335&ref=testingcatalog.com) focused on each company’s next move. This dynamic underscores the intensifying rivalry at the close of the year, with both Google and OpenAI pushing to set new benchmarks in model performance.