Best Multimodal Models for Antigravity CLI

Gemini 4 Argon is a frontier AI model from Google DeepMind built to sustain deep reasoning across complex, long-running professional workflows. Google designed the model for demanding work spanning software engineering, finance, legal tasks, enterprise knowledge work, cybersecurity defense, and creative writing. Argon supports coding, reasoning, multimodality, and multi-step task execution, allowing it to work across workflows that require information gathering, analysis, tool use, and extended problem solving. Its output token limit has been increased from 64,000 to 1 million tokens, giving the model additional capacity for lengthy reasoning and generation within a single trajectory. On DeepSWE v1.1, Google reports a score of 77.9% for real-world long-horizon software engineering, while its AutomationBench score of 51.3% measures performance on end-to-end business workflows. Google also reports strong results on evaluations covering finance, legal work, visual analysis, and long-video understanding, including a 91.7% score on LVBench. For cybersecurity teams, Argon can autonomously discover, validate, and patch software vulnerabilities and achieved a reported 68% score on CWE-bench v1. Google is initially providing the model to selected cyber defenders through its Fairwind Program while strengthening safeguards before expanding access to developers, enterprises, and consumers. Argon is planned to launch at an introductory price of $2 per million input tokens and $10 per million output tokens, with cached input tokens receiving a 95% discount from the standard input price.

Gemini 4 Carbon is an experimental AI model reportedly being developed by Google DeepMind for advanced reasoning and software engineering applications. The model has appeared in reports describing internal testing of multiple Gemini 4 variants, including Argon, Barium, and Carbon. Carbon has reportedly been integrated into Google's internal Jetski coding environment for evaluation by company employees. Early internal feedback has highlighted its potential performance in complex coding tasks, although independently verified benchmark results have not been published. Its reported development suggests an emphasis on improving AI-assisted programming and technical problem-solving capabilities. Potential use cases include code generation, debugging, codebase analysis, software maintenance, and autonomous development workflows. Carbon belongs to the broader Gemini 4 research effort, which includes publicly released models supporting multimodal reasoning, software engineering, and AI agent applications. However, Google has not confirmed whether Carbon will become a standalone product, an update to Gemini 4 Argon, or a different publicly branded model. Technical specifications, API availability, pricing, deployment options, and release timing remain unannounced.