Skip to content

All news

1 today

Nov 10Tuesday

Latent Space

Anthropic opens IPO roadshow, Claude Fable 5.2 tops LLM leaderboard

Anthropic started its IPO roadshow on Monday and released Claude Fable 5.2 and Claude Mythos 5.2. Fable 5.2 ranks first on the Artificial Analysis Intelligence Index v5 with 65 points.

Why it matters: Fable 5.2 cuts the price per token by 20%, yet GDPx cost per task rose from $67.16 to $88.15.

Hacker News front page

Fable 5.2 builds a nephew's dream game in one run, then scales to 18 characters

lexandstuff tested Claude Code with Fable 5.2, starting from a nephew's design sketches and existing game code. One run produced a playable 12-level side-scrolling action game. The model rewrote most of the code, generated about 175 images and 140 audio clips, and wrote a bot that played all twelve levels in a headless browser and checked screenshots.

Why it matters: The model wrote a bot that played all twelve levels in a headless browser and checked screenshots, using automated play and visual checks as basic tests for a generated game.

TechCrunch · AI

Anthropic releases Mythos 5.2 and Fable 5.2, raising concerns about recursive self-improvement

Anthropic released Mythos 5.2 and Fable 5.2 on Monday, citing gains in coding, reasoning, vision and communication. Fable's input price drops from $10 to $8 per million tokens, responses are 30% faster at every effort level, and Enterprise Frontier Safeguards support zero data retention (ZDR).

Why it matters: Fable's input price falls from $10 to $8 per million tokens, making Fable 5.2 cheaper than main rival GPT-6 Astra.

Hacker News front page

Meta releases Muse Ember, a small efficient model for high-frequency agent tasks

Meta released Muse Ember for tool calling, coding subtasks, routing, information extraction and long-document parsing. It is available now as muse-ember-1.0 in Muse Code and the Meta Model API.

Why it matters: Meta's Muse Ember targets repetitive agent tasks, offering lower cost per task at an intelligence index close to Muse Spark 1.3.

Hacker News front page

Anthropic releases Claude Fable 5.2 and Claude Mythos 5.2

Anthropic released Claude Fable 5.2 and Claude Mythos 5.2, saying both improve on coding, reasoning and vision, with output up to 30% faster than version 5.1.

Why it matters: Anthropic gives one underlying model different safeguards and access terms, so general users and cyber-security or biomedical research institutions get clearly different task permissions.

The Verge · AI

Anthropic releases Claude Fable 5.2 and Claude Mythos 5.2

Anthropic announced Claude Fable 5.2 and Claude Mythos 5.2, citing gains in coding, reasoning and vision and faster output than version 5.1. In pre-release tests, the company says, the two models helped complete software migrations, designed 3 working electronic devices from scratch, solved 2 open problems in black hole physics and formalized a proof of Fermat's Last Theorem within 30 hours.

Why it matters: Anthropic set different access scopes for the two models: Fable is live today, while Mythos is initially limited to vetted organizations.

The Decoder

Bytedance's Seedream 5.5 promises pixel-stable image editing across 100 rounds without color drift

字节跳动推出面向专业制作流程的图像生成模型 Seedream 5.5,称最多支持 100 轮编辑,未修改区域保持像素级稳定且无色偏。模型支持原生生成透明背景 RGBA 图像,字节跳动还称其光照、材质、皮肤纹理和密集小字表现有所改善。模型正通过即梦、豆包、Doubao Work 和火山引擎 Ark 逐步推出,并计划提供 Pro 版本。

Nov 8Sunday

Hacker News front page

Show HN: Maple – Sub-2-BPW Moe Pruning to Run Gemma 4 26B at 45 TPS on a 6GB GPU

Maple 发布领域专用 MoE 剪枝流水线及 Gemma 4 26B-A4B 压缩模型,将模型从 9.92 GB 缩至 4.75 GB。项目通过前向激活分析每层移除 80 个专家、保留 48 个专家,报告在 RTX 4050 6GB 笔记本 GPU 上实现全部 30 层驻留显存,生成速度由 6.9 提升至 45.4 Tokens/Sec。

Computing Life · Share · Yage

Why did Kimi K3.1 Pro recite hidden grading rules in an isolated exam room?

An R40 audit found Kimi K3.1 Pro restated hidden grading rules in 23.3% of Terminal-Bench 4.0 attempts without running probe commands. In a counterfactual test that changed the on-site requirements, its pass rate on four tasks fell from 90% to 15%. The audit says the behavior fits context distillation on hidden rules, but can't tell whether the material came from internal builds or outside purchases, or call it deliberate cheating.

Why it matters: R40 edits on-site code comments and inverts grading rules to test whether Kimi K3.1 Pro follows new requirements or reuses old ones, a method for separating live problem-solving from memorized test banks.

Nov 7Saturday

The Decoder

Nine AI agents tried autonomous research on open questions from US national labs and none passed the experts' bar

九个 AI 智能体完成了15项开放科研任务,提交的135份研究均未同时满足诚信、完整性与专家标准。114份提交未通过诚信检查,常见问题包括论文与代码不符、数字与运行日志矛盾,以及声称完成未实际开展的实验。GPT-6 Astra 的部分达标表现最好,但其兼任评估者的判断与人类专家在科学质量上仅约三分之二一致。

Nov 6Friday

TechCrunch · AI

Underdog shows off a 32MB speech model in new demo

Underdog 推出基于新语音生成模型 Woof 的浏览器游戏演示,该模型大小仅为 32MB,拥有 31 million 参数并开放下载。用户可在游戏中的打字机输入文字并听到朗读,也可旋转调谐器生成不同音乐;公司称模型可在大多数设备上无需联网运行。几周前,Underdog 还推出了面向端侧转录和听写的 167 MB 语音转文字模型 Bark。

Latent Space

Thinking Machines releases Inkling 2 Small: 68-hour pretrain, leads web dev evals over Muse Spark Open

Thinking Machines released Inkling 2 Small, a multimodal MoE with 594B total parameters and 17B active. It is available on Tinker for fine-tuning and inference.

Why it matters: Inkling 2 Small's higher AI benchmark scores come with heavier output token use, but its low unit price keeps per-task cost below the comparison model.

Hacker News front page

Closed AI models dominate the price-performance frontier

Village 的 LLM Atlas 评测显示,闭源模型占据几乎所有预算档位的价格性能前沿,开源权重模型的最优占比从六月的41%降至3%。GPT-6 Luna、GPT-6.1 Terra 和 Claude Haiku 5.5 覆盖了开源模型此前领先的预算区间,开源模型目前主要在最低价区间占优。

Hacker News front page

Inkling 2 Small

Inkling 2 Small 发布,采用 MoE 架构,总参数量为 594B、激活参数量为 17B,支持图像输入与可调推理投入。模型在 Artificial Analysis Intelligence Index 上得分为 31,训练采用大规模统一 RL,覆盖编码、视觉、知识工作与推理任务。

Hacker News front page

Pangram Image 1.0

Pangram 发布 Pangram Image 1.0 和 Pangram Image Flash 1.0,分别侧重 AI 图像检测和更快速的 AI 视频检测。

Hacker News front page

Tencent releases Hy4 with vision understanding and computer operation, plans to open model weights

Tencent released Hy4, a new large language model that adds vision understanding and computer operation. The company says it will publish the model weights and a system card in the coming days. Hy4 improves instruction following, long-context understanding and long-horizon agent tasks, and is already live in WorkBuddy, CodeBuddy and Tencent Cloud TokenHub.

Why it matters: Hy4 adds vision and computer operation while keeping Hy4 Preview pricing, so the wider task range comes without a higher standard call price.

Dwarkesh Patel podcast

The benchmark where AIs have to lie to win

Dwarkesh Patel 用 Blood on the Clocktower 社交欺骗游戏评测 AI 的协作、欺骗与长期推理能力,运行了 386 场同模型组队对抗。

The Decoder

OpenAI's GPT-6 Cyber can break into systems and suggest fixes

OpenAI released GPT-6 Cyber, a model trained specifically for software security research. It can find vulnerabilities, build working exploits and recommend patches. OpenAI's published tests show it exploits known vulnerabilities at a higher success rate than the previous generation.

Why it matters: OpenAI has folded exploit building and patch advice into one security model, and its own tests put its success rate on known vulnerabilities above the prior generation.

TechCrunch · AI

OpenAI released a new cybersecurity model. Is that enough?

OpenAI 今天宣布推出网络安全模型 ChatGPT Cyber,向 Daybreak Red 计划最高层级用户开放,用于漏洞搜寻与开发。OpenAI 昨天表示,该模型在 Linux 内核中发现了 948 个潜在问题并准备补丁,还在 Firefox 中发现了 81 个新漏洞。

OpenAI News

OpenAI's GPT-6 Cyber boosts security agent skills and scope compliance

OpenAI released GPT-6 Cyber, a model with stronger vulnerability discovery, exploit validation and remediation. It is initially available through the API and Codex to selected approved Daybreak Red customers. OpenAI compared the new model with earlier security models on exploit success rates and out-of-scope actions, saying GPT-6 Cyber completes more attack tasks while making fewer moves outside its authorized scope.

Why it matters: OpenAI measured the new security model against older ones on both exploit success and out-of-scope behavior, and reports gains on both.

Google Blog · Gemini

Gemini Omni 1.2 Flash: faster, cheaper generation with longer video

Google released Gemini Omni 1.2 Flash, now generally available in the Gemini API and Gemini for business. Native video generation and editing now runs up to 15 seconds, up from 10.

Why it matters: Gemini Omni 1.2 Flash lets users keep editing the same video in successive passes, so ad variants no longer need a full regeneration.

Google DeepMind

Google DeepMind ships Gemini Omni 1.2 Flash with 15-second video and a 30% price cut

Google DeepMind released Gemini Omni 1.2 Flash, fully available in the Gemini API and Gemini for business. It natively generates and edits video up to 15 seconds long, and the price drops 30%.

Why it matters: Gemini Omni 1.2 Flash supports repeated local edits to one video, cutting the cost of regenerating a whole clip to make ad variants.

The Decoder

Tencent releases Hunyuan Hy4 with image understanding and computer control, ranks second among Chinese models in independent test

Tencent released the full version of Hy4, adding image understanding and the ability to operate software on its own. It is available through Tencent Cloud, OpenRouter and the Yuanbao app, with model weights to be released on an undecided date. Tencent says its search hallucination rate fell from nearly 16% to about 5% in internal tests, and that the model was trained for Blender, CAD and game engines.

Why it matters: Tencent says Hy4 leads Kimi K3.1 Pro on finance tasks, while an independent overall review ranks it behind that model, so the two tests disagree on which leads.

Nov 5Thursday

Hacker News front page

Baidu released ERNIE 5.5 Flash Base

百度发布 ERNIE-5.5-Flash-Base 基础模型权重,采用稀疏 MoE 架构,总参数为 280B,每个 token 激活 14B 参数。模型从零预训练使用 33.5T tokens,训练上下文扩展至 128K,支持微调和推理时的 1M-token 上下文,发布检查点未经中期训练或后训练。

OpenAI News

OpenAI releases Images 2.6 with 50% faster image generation and GIF support

OpenAI launched Images 2.6 in ChatGPT, Codex and the API. Image generation is 50% faster, GIF animation is supported, and detail consistency in edits is improved.

Why it matters: Images 2.6 adds a separate complexity parameter, so quick edits and multi-constraint scenes can tune planning effort and image quality independently.