Users report receiving Gemini 4 Argon outputs through Google A/B tests
Outputs shared on X look like a clear step up from Gemini 3.8 Flash, but Google has not opened the model to the public.
Model releases, capabilities, and platform changes from the frontier labs and open ecosystems.
Outputs shared on X look like a clear step up from Gemini 3.8 Flash, but Google has not opened the model to the public.
The model costs $0.042 per million input tokens, output is free, and it is available through Microsoft Foundry and OpenRouter.
The California company says the model scores at the top of physics benchmarks and is meant to serve as a base for controlling robots, drones and game characters.
References to Gemini 4 Argon with context options of up to 900K tokens have also appeared in Google's Antigravity tool ahead of a wider release.
In Anthropic's own test, the new dynamic workflows found 66 of 70 hidden bugs, while a single agent found 14 to 27 per run.
Google is releasing it to businesses first and says consumers will come later. Sundar Pichai says Gemini has more than 1 billion monthly users.
GPT-6 Sol and GPT-6 Luna replace GPT-5.6 in ChatGPT on free and paid plans. Answers can now include charts, buttons and small tools that users can work with directly.
The small model is pitched for high-volume work and as a subagent alongside Opus 5.5 and Sonnet 5.5. Max and Team subscribers also get a new monthly API credit.
The French company says the multimodal model beats every open-weight model from the US or Europe. Until the release, it is being red-teamed with security partners and state authorities.
A 1K image now costs 3.36 cents, down from 6.70 cents, and the model can work from up to 14 reference images at once.
Weights are due under Apache 2.0 later this month. Beam trails DeepSeek and Kimi on coding tests, and its efficiency claim rests on a rough formula.
The 78-billion-parameter mixture-of-experts model was trained in Germany and Finland and is available under an Apache 2.0 license.