1
Fable 5. Safety Taken to an Extreme
测试Fable 5时因安全机制过严被自动降级至弱模型，暴露AI安全策略与用户需求的真实冲突
I finally decided to try out Fable 5 using the standard Claude.ai interface. I have a go-to test prompt that answers simple kids' questions in an absu…
2
网友反馈 Claude Fable 5 重上架版“降智”，严苛 AI 护栏致频繁回退 Opus 4.8
Claude Fable 5重上架被用户吐槽“降智”，严苛护栏致频繁回退旧版Opus 4.8
IT之家 7 月 4 日消息，科技媒体 bleepingcomputer 昨日（7 月 3 日）发布博文，报道称 Anthropic 本月重新上架 Claude Fable 5 最强模型， 除了额度限制外，多名用户反馈存在“降智”情况，且更频繁回退到 Opus 4.8 模型。 Claude Fabl…
3
A Red-Team Study of Anthropic Fable 5 & Opus 4.8 Models
最新Anthropic模型红队安全测试结果，揭露Fable 5与Opus 4.8的脆弱性。
arXiv:2606.18193v1 Announce Type: cross Abstract: We evaluate the adversarial robustness of two frontier large language models (LLMs) developed by Ant…
4
Claude Fable 5 Scores 95% on SWE-bench, Then Hands Off to Opus 4.8
Anthropic全新Claude Fable 5在SWE-bench验证集上高达95%，并顺手交接给Opus 4.8，AI模型性能再刷新纪录。
The headline number is 95% on SWE-bench Verified. That's the score attached to Claude Fable 5, Anthropic's new general-access model in the Mythos clas…
5
Fable 5自带反蒸馏机制！检测到就降智，误触率高到离谱
Fable 5内置反蒸馏检测，误触率高到离谱，聊个天都可能被降智切换老模型。
Claude刚刚发布的新模型Fable 5，很多人可能压根就用不上！
6
Show HN: Built a browser game inspired by Rust
一款受《Rust》启发的浏览器生存游戏，用AI模型Opus 4.8花一个周末建成，开源可玩。
Wanted to see how far I could get with Opus 4.8 and was impressed. Got tripped up in a few places with AI game behavior, but eventually got it to a go…
7
Anthropic leapfrogs OpenAI as most valuable AI startup
Anthropic估值9650亿美元超越OpenAI，并发布Opus 4.8模型，AI初创格局生变。
Article URL: https://www.businessinsider.com/anthropic-surpasses-openai-with-965b-valuation-debuts-opus-4-8-2026-5 Comments URL: https://news.ycombina…
8
llm-anthropic 0.25.1
llm-anthropic 0.25.1 更新：默认 max_tokens 改为各模型最大输出，不再硬限制 8192，适配 Claude Opus 4.8 等新模型
Release: llm-anthropic 0.25.1 New model: Claude Opus 4.8 ( claude-opus-4.8 ). New -o fast 1 option for fast mode , for organizations with that feature…