Google unveils Gemini 4 Argon at half price
Photo: Collected
Google has launched Gemini 4 Argon, a new artificial intelligence model engineered for coding, research, and complex problem-solving, making it available to developers and businesses on Tuesday at introductory prices set at half the standard rates.
Developers can integrate the model into their software through the Gemini API, while Google AI subscribers can access it via Antigravity, the Gemini app and AI Mode in Google Search. Initially, input tokens will cost $2 per million and output tokens $10 per million, with a 95 per cent discount on previously processed, cached input tokens. However, this promotional pricing will remain in force only until November 30, 2026, after which the rates will double to $4 and $20 per million tokens, respectively. Tokens, it should be noted, are the basic units through which AI models process information.
Google claims that Gemini 4 Argon can tackle sophisticated tasks in coding, economics, law, cybersecurity and scientific research, outperforming other leading models in several capability assessments. Furthermore, the company said the model ranked first in the Vals Index, which gauges performance across economically significant sectors. Its integration into Antigravity has reportedly raised AI agents’ code submission rates by more than 35 per cent while cutting the need to revert code changes by roughly a third. Google also said the model helped disentangle core webpage functionality from legacy Chrome browser code, compressing work that would ordinarily take months into just a few weeks.
Read More: Meta launches Muse Wonder models to strengthen AI capabilities
According to Google, several organisations, including Deloitte, LexisNexis, Ambience Healthcare and Isomorphic Labs, are testing the model for research, document synthesis, long-term information analysis and large-scale codebase analysis. Moreover, the company reported that integrating Argon into its Agent Development Kit reduced session preparation time by 40 per cent, while molecular structure prediction in XTB software became up to 2.3 times faster. Google is also expanding its Fairwind programme by bringing in several hundred additional partners. Collectively, these initiatives underscore the company’s broader ambition to extend AI adoption across software development, scientific research and business operations.
newsbangladesh.com/mb








