PulseAugur
中
实时 11:10:56
English(EN) Ethan Mollick @emollick.bsky.social If this holds up, it is big: "We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math. It directly changed only 0.

AI模型泄露暗示 Claude Mythos 5.5、GPT-6 Astra 和 Gemini 4

作者 PulseAugur 编辑部 · [3 个来源] ·

最近的一份泄露表明，Anthropic 的 Claude Mythos 5.5（代号 Fable 5.5）将是一项重大进展。此外，传言指向即将推出的 GPT-6 Astra 和 Gemini 4 Argon Xenon 等模型，这表明主要 AI 实验室的开发速度很快。Ethan Mollick 强调了一个潜在的突破，即 GPT-6 Astra 被用来编辑 Qwen3.5-0.8B 的权重，在没有传统训练方法的情况下提高了其数学和指令遵循能力。 AI

影响 模型编辑和推理能力的潜在进步可能会加速 AI 的开发和性能。

排序理由 该集群由讨论泄露和潜在 AI 模型进展的社交媒体帖子组成，而非官方公告。

在 Mastodon — fosstodon.org 阅读 →

AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →

AI模型泄露暗示 Claude Mythos 5.5、GPT-6 Astra 和 Gemini 4

本文如何被排名

Signal score
0 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Commentary
该集群由讨论泄露和潜在 AI 模型进展的社交媒体帖子组成，而非官方公告。
Source corroboration
3 independent sources
Multiple independent publishers reporting the same story raises confidence that it's real and newsworthy.
Topics
model release, product
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Same-day
Cluster formed today. Ranking reflects the current source set at time of score.
Coverage growth since scoring
+1 source(s) since last score
New sources have picked up this story since our last re-score. Score will update on the next scoring pass.

完整方法见我们的编辑标准。

报道来源 [3]

  1. Mastodon — fosstodon.org TIER_1 English(EN) · [email protected] ·

    @ WorldofAI on YT! HUGE CLAUDE # MYTHOS 5.5 LEAK! # Fable5 .5 INSANE, # GPT -6.2 Astra, # Gemini4 Argon Xenon & Carbon-H! # AI NEWS 📰 https:// youtu.be/L7opnQdp

    @ WorldofAI on YT! HUGE CLAUDE # MYTHOS 5.5 LEAK! # Fable5 .5 INSANE, # GPT -6.2 Astra, # Gemini4 Argon Xenon & Carbon-H! # AI NEWS 📰 https:// youtu.be/L7opnQdpdMY?si=qvUvFZ ug-_zxciF

    链接 youtube.com/watch youtube.com/watch
  2. Mastodon — fosstodon.org TIER_1 English(EN) · [email protected] ·

    Ethan Mollick @emollick.bsky.social If this holds up, it is big: "We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math. It directly changed only 0.

    Ethan Mollick @emollick.bsky.social If this holds up, it is big: "We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math. It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without grad…

    链接 rikiyat.github.io/…/weight-space-programm…
  3. Mastodon — mastodon.social TIER_1 English(EN) · hnjsdev ·

    🏆 Best of Hacker News — Nov 2, 2026 Astra can read and write weights to improve Qwen 2531 points · 782 comments 📖 HN Reader: https:// hn.jsdev.space/story/50341

    🏆 Best of Hacker News — Nov 2, 2026 Astra can read and write weights to improve Qwen 2531 points · 782 comments 📖 HN Reader: https:// hn.jsdev.space/story/50341174 🔗 Original: https:// rikiyat.github.io/blog/weight- space-programming/ 💬 HN: https:// news.ycombinator.com/item?id=5…

    链接 rikiyat.github.io/…/weight-space-programm…

相关实体

相关话题