Google has launched its Gemini 4 Argon model with an introductory price of $2 per million input tokens and $10 per million output tokens. However, independent evaluations suggest that while the model performs competitively on benchmarks, it consumes roughly twice as many tokens per task compared to similar models. This increased token usage means that the actual cost per task could be significantly higher than initially advertised, especially when the introductory pricing expires and the cost doubles to $4 input and $20 output per million tokens. Furthermore, the model is not yet widely available, with initial access restricted to a cybersecurity defender program. AI
IMPACT Higher token consumption per task suggests that actual operational costs for Gemini 4 Argon may significantly exceed introductory pricing, impacting budget planning for AI deployments.
RANK_REASON Frontier-lab model release with system card and pricing details. [lever_c_demoted from frontier_release: ic=2 ai=1.0]
- Artificial Analysis
- Claude Opus 5.5
- Fairwind Program
- FrontierSWE v2
- Gemini 3.1-pro-preview
- Gemini 4 Argon
- GPT-6 Astra
- GPT-6 Sol
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →