实测:Gemini 4 Argon 难守字数上限,Opus 5.5 安全拦截过频
jessi_cata · x · 2026-11-09
A user shared real-world eval logs of instruction-following failures:
- Gemini 3.8 Flash and Gemini 4 Argon repeatedly ignore output length limits given in the prompt, producing essays far past the cutoff — the attached table shows dozens of essaysizelimit failures, some with 10k+ output tokens against a 16k limit.
- Claude Opus 5.5 frequently hit refusal at the prisoner stage; the author says its cyber safeguards are over-active.
- The table also shows Qwen3.8 Max Prime failing exactly at its 6000 outputtokenlimit.
Net takeaway: models differ sharply at constraint-following — Gemini can't stay brief, Opus blocks too eagerly.
「模型」频道最新
- 开发者放话要用 Qwen 4 实测,阿里新模型待检验 — jasonkneen · 2026-11-09
- 安全研究员用 GLM 批量把 Windows 驱动变提权 0day,追平 Opus — MalwareTechBlog · 2026-11-09
- Nous Research Arete-27B 现已提供 GGUF 量化版本 — gladkos · 2026-11-09
- 安全研究者 MalwareTech：GLM 5.3 是我性价比首选，abliterated 版反而更差 — MalwareTechBlog · 2026-11-09
- 安全研究员实测:GLM 5.3 护栏宽松,可直接用于漏洞挖掘 — MalwareTechBlog · 2026-11-09
- OpenAI 自研自动化报告：模型 80% 时间跨度仅 15 分钟 — tobyordoxford · 2026-11-09