gemini-4POST
Gemini 4 Argon 宣称 DeepSWE 77.9% 但被指编程仍吃力
Google 称 Gemini 4 Argon 在 DeepSWE 拿到 77.9%、支持 100 万输出 token，内部已用其做 C++ 转 Rust；但彭博消息称它部分编程任务仍吃力。
Sad Creator
@SadCreatorTalks25 天前收录原推发布于 2026.10.04
Sad Creator @SadCreatorTalks原推发布于 2026.10.04
Gemini 4 Argon: Google claims 77.9% on DeepSWE (above Opus 5.5 and GPT-6 Astra) plus 1M output tokens. Internal agents already doing C++ to Rust, even Fuchsia kernel work. Bloomberg: people with access say it still struggles on some coding tasks. I believe both. That's the problem.
x.com/sadcreatortalks/status/2106768944774500367采集于 2026.10.05 09:58