# TestingCatalog — Gemini 4 Argon spotted in A/B testing on Antigravity

- Company: TestingCatalog (testingcatalog.com)
- Announced: 2026-10-11T21:58:46+00:00
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://www.testingcatalog.com/gemini-4-argon-ab-tests-early-outputs/
- Record: https://forck.live/items/19310-gemini-4-argon-spotted-in-a-b-testing-on-antigravity
- Subject: AI models / agents / unreleased features
- Models affected: Gemini 4 Argon, Gemini 3.8 Flash, Gemini 3.1 Pro
- Context window: 256K, 512K, and 900K

Google's unreleased Gemini 4 Argon model has been spotted in A/B tests, with some users on X sharing outputs they believe come from the model. Google claims a 77.9% score on DeepSWE and first place on the Vals Index for finance, legal, and tax work. The company is preparing Antigravity for Argon with Ultra upgrade prompts and Argon-only context options of 256K, 512K, and 900K.

## Evidence

Verbatim from https://www.testingcatalog.com/gemini-4-argon-ab-tests-early-outputs/:

> Google initially announced Gemini 4 Argon on September 30. Since then, the company has been restricting access to a select group of users. The company claims a state-of-the-art 77.9% on DeepSWE, a test of long software engineering tasks, along with first place on the Vals Index for finance, legal, and tax work and on Zapier's AutomationBench.

---

Record: https://forck.live/items/19310-gemini-4-argon-spotted-in-a-b-testing-on-antigravity
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md