Google's Gemini 4 Argon Aims to Outperform Rivals in AI Market

Alphabet's shares rose around 2% in premarket trading on Thursday after Google announced the launch of its new artificial intelligence model, Gemini 4 Argon. The company claims that this latest model has stronger coding and cybersecurity performance than rival models at lower prices.

The initial pricing for Gemini 4 Argon is $2 per million input tokens and $10 per million output tokens, which Jefferies analysts say is well below comparable offerings from Anthropic. Claude Fable 5.1 costs $10 and $50 respectively, while Claude Opus 5.5 is priced at $4 and $20.

Google's Gemini 4 Argon scored a record result of 77.9% on the DeepSWE v1.1 coding benchmark, according to Jefferies. The model ranked first for coding on LLM Arena's Text Arena and eighth on Code Arena: Web Dev, 21 places higher than its predecessor, Gemini 3.8 Flash.

Jefferies noted that while Argon trails leading models on the more demanding FrontierSWE v2 test, it tied Grok 4.7 for the top score of 68% on CWE-bench v1, which tests vulnerability fixes. Google also said that its model can identify, verify and fix software vulnerabilities autonomously.