Google Pushes the Boundaries of AI with Gemini 4 Argon
Google announced its new AI model, Gemini 4 Argon. We examined the test results, pricing, and impact on the sector of the model, which competes with GPT-6 Astra.
Google recently officially announced its highly anticipated new artificial intelligence model, Gemini 4 Argon. This model, which has caused a great stir in the technology world, is leaving its competitors behind, particularly in complex workflows, cyber defense, and software development. Currently undergoing testing by the US government and with limited widespread access, Gemini 4 Argon is directly competing with OpenAI’s GPT-6 Astra model in benchmark tests. This performance increase, which confirms Google’s internal RSI (Relative Strength Index) rumors, signals the beginning of a new era in AI technologies.
- Gemini 4 Argon leads in 14 out of 19 key AI benchmark tests.
- The new model has achieved one of the safest AI models in its category with a 15% hallucination rate.
- The model surpassed leading models in the sector with a 77.9% success rate in the DeepSWE v1.1 test.
- Google is offering a special 50% discount on the Gemini 4 Argon at the market entry level.
Gemini 4 Argon Takes the Lead with Test Results
According to information shared by Artificial Analysis, Gemini 4 Argon has made a significant leap, especially in agent evaluations. Offering valuable improvements in areas where previous Gemini versions lagged behind, the model scored 77.5% on AutomationBench-AA, surpassing its competitors.
The system also holds a strong position in demanding tests such as Terminal Bench 4, and demonstrates a consistent profile in Analytical Quality and Presentation Quality criteria.
The Model’s Performance and Cost Balance is Noteworthy
Gemini 4 Argon is Google’s first non-Flash model after a seven-month hiatus. Reaching the same level as GPT-6 Astra with a score of 53 in the AI index, the model also follows a highly competitive pricing strategy. Initially offered at a $2 entry and $10 exit price per million tokens, the model aims to remain at market averages even after the discount period ends.
Discussions and Criticisms Continue in the Sector
Although the test results are impressive, there are some question marks regarding the model’s real-world performance. A report published by Bloomberg stated that some employees within Google expressed concerns that the model was struggling to fulfill its coding missions. This shows that academic successes may not always translate fully into practical applications.
How much do you think Google’s new Gemini 4 Argon model will change the dynamics of the AI world? In its competition with GPT-6 Astra, which do you think has the advantage? Don’t forget to share your opinions and thoughts on the subject with us below.
