Artificial Intelligence
Google launches Gemini 4 Argon
Google released Gemini 4 Argon and said the model outperformed several rival systems in benchmarks, began rolling it out to paid users, and set introductory token pricing below competitors.
- Google launched Gemini 4 Argon.
- Google said Argon offered better reasoning and improved performance on complex workflows.
- Google said Argon beat GPT-6 Astra, Claude Fable 5.1 and Claude Opus 5.5 across benchmarks in coding, cybersecurity, science, finance and law.
- Google said Argon was already being used internally to find security flaws, reduce the computing power needed for YouTube comments and debug Wing’s drone physics simulator.
- Google said the model started rolling out to Pro and Ultra subscribers in the Gemini app and AI mode in Search, and to Ultra subscribers on Antigravity.
Google launched Gemini 4 Argon and said it was meant to compete with other frontier AI systems. The company said the model improved reasoning and handled complex workflows better than its earlier releases.
Google said Argon outperformed GPT-6 Astra, Claude Fable 5.1 and Claude Opus 5.5 in benchmark tests covering coding, cybersecurity, science, finance and law. It also said the model was already being used inside the company to find and patch vulnerabilities, lower the computing power needed for YouTube comments and debug Wing’s drone simulator.
The rollout began for higher-tier Pro and Ultra subscribers in the Gemini app and AI mode in Search, and for Ultra subscribers on Antigravity. Google said it was not yet clear whether Plus subscribers or free users would get access.
Google set introductory prices at $2 per million input tokens and $10 per million output tokens, with both rates due to double at the end of November. The launch followed a limited release to selected trusted cyber defenders for risk testing, and came after Gemini 3 in late 2025.
Named in this story
Companies
- launched Gemini 4 Argon and made the performance and pricing claims
- Alphabet
- was named as Wing’s parent company
- Wing
- had its drone physics simulator cited as an internal debugging use case
- OpenAI
- was named as a competitor in Google’s pricing comparison
- Anthropic
- was named as a competitor in Google’s pricing comparison
- Meta
- was cited as having announced a frontier model the day before
Products and systems
- Gemini 4 Argon
- the model Google launched and rolled out
- GPT-6 Astra
- one of the rival models Google said Argon beat
- Claude Fable 5.1
- one of the rival models Google said Argon beat
- Claude Opus 5.5
- one of the rival models Google said Argon beat
- YouTube
- had its comments section cited by Google as an internal use case for Argon
- Gemini app
- one of the products where Argon began rolling out
- Antigravity
- the platform where Ultra subscribers could use Argon
- Muse Wonder Full
- Meta’s frontier model mentioned for comparison
How the source tells it
The piece was promotional and race-driven, mixing Google’s claims with an editorial comeback narrative.
- vendor boosterism through repeated presentation of Google’s own benchmark, pricing and rollout claims as proof of capability
- competitive urgency through frontier-race language and repeated emphasis on rivals, timing and market position
- underdog comeback through a turnaround frame that cast the launch as evidence Google was back in contention