Google has announced Gemini 4 Argon, a new large language model designed for long-horizon tasks, particularly in software engineering, enterprise knowledge work, and cybersecurity. The model is being released in a phased approach, initially to a select group of cyber defenders through its Fairwind Program, with broader access planned for paid API customers and Google AI Ultra subscribers. Gemini 4 Argon boasts impressive benchmark scores on various evaluations, including DeepSWE v1.1 and LVBench, and features a significant increase in its output token limit to 1 million. AI
IMPACT This release pushes the boundaries for long-context AI tasks, potentially accelerating enterprise adoption in complex domains like cybersecurity and software engineering.
RANK_REASON Frontier-lab model release with gated access and benchmark numbers. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
- CWE-bench v1
- DeepSWE v1.1
- Fairwind Program
- Gemini 3.8 Flash Cyber
- Gemini 4 Argon
- LVBench
- Zapier's AutomationBench
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →