gemini-4POST
Gemini 4 Argon 单次可生成 100 万 token
DeepMind 发布旗舰模型 Argon，单次响应上限从 6.4 万提升到 100 万 token，DeepSWE v1.1 得 77.9%，主打长周期软件工程与企业研究。
Hải Ly
@leslieMen28 天前收录原推发布于 2026.10.01
Hải Ly @leslieMen原推发布于 2026.10.01
Google DeepMind launched Gemini 4 Argon. The flagship model can generate up to 1 million tokens in a single response, up from 64,000 in previous Gemini models. Built for long-horizon software engineering, enterprise research and cybersecurity. Argon scored 77.9% on DeepSWE v1.1, beating Claude Opus 5.5 and GPT-6 Astra. The biggest shift is not the benchmark score. It is the ability to reason, use tools and complete massive workflows without losing context. 🤖
x.com/lesliemen/status/2105474737891848405采集于 2026.10.02 10:15