Gemini 4 Argon now available to the public: Google’s response to ChatGPT-6 and Fable 5
Following the preview for Fairwind partners, Google is opening up access to developers and Pro users for its new frontier model. Ranked number one in the Vals Index for finance, coding and security, it is launching with promotional prices until 30 November and is focusing on agent-based development in Antigravity.
Unveiled less than two weeks ago, when it was made available exclusively to partners of the Fairwind cybersecurity project, Gemini 4 Argon is now being released to the public. Or rather, developers and businesses will be able to access it via the APIs, as will Google AI Pro subscribers via the Gemini app and the AI Mode search feature, whilst Ultra subscribers will also be able to use it within Antigravity, Google’s agent-based development platform.
Gemini’s new model ranks first in the Vals Index, a tool that assesses the capabilities of large language models (LLMs) when applied to finance, software development and cyber security, outperforming the other models available on the market. Upon its release, Google also clarified the pricing: until 30 November, it will cost $2 per million input tokens and $10 per million output tokens. From 1 December, these figures – expressed in dollars – will double.
In the Gemini app, Argon promises to be more direct in its responses, to avoid repetition and to steer clear of clichés. For longer tasks, it will keep the user updated on their progress more frequently than previous models do. What’s more, the app can connect with over 40 applications, such as Instacart, OpenTable, Adobe and Shopify, enabling users to carry out tasks ranging from ordering groceries to growing their business.
As for Antigravity, Big G’s new model promises greater speed and precision: used by over 100,000 of the company’s employees, in its first four weeks of use it recorded a 35 per cent increase in the amount of code developed by agents, whilst reducing code rollback – that is, the need to undo changes to revert to previous, stable versions – by 30 per cent.
Announced in May at Google I/O, Google’s developer conference, and originally scheduled for release in June, Argon 4 is arriving a few months late, after both Anthropic and OpenAI have released their new state-of-the-art models, Fable 5 and ChatGPT-6. It remains to be seen whether and how the performance of the Mountain View-based company’s new LLM – which is said to be superior to that of its competitors – will reshape market shares amongst the leading Western players in artificial intelligence.
