Someone shared how a solo dev builds a Ragnarok Online x Monster Hunter-style game with Claude Code on a Max 20x plan. His three custom skills: → route: sends easier tasks to Sonnet or Haiku, with Opus reviewing → handoff: starts a fresh session with the context needed to continue → Re-searcher: saves research findings to Obsidian, so notes don't become summaries of summaries The stack: Godot, Tripo P2 for 3D assets (about $20/month), Blender, and prompts saved as JSON for a consistent art style. Caveats: the poster is relaying what the dev told them, and the game is still marked early development. Reddit: reddit.com/r/ClaudeAI/commen…

Kimi K3.1 Pro is among the leading models in agentic knowledge work and coding, but early signs suggest it will rank lower on AA-3D We are still collecting the preference votes for its Elo on AA-3D, our benchmark for how well AI models build interactive 3D scenes. In agentic knowledge work and coding, it is among the leading models on GDPval-AA and Terminal-Bench 4.0. Its predecessor, Kimi K3, is 12th of 22 models on AA-3D. Here is one of its AA-3D scenes: a 492 m mountain-bike descent with berms, roots, a rock garden, a drop, and a jump. A demo run rides it from top to bottom, and you can take over the steering, pedaling, and braking at any point. Congratulations to @Kimi_Moonshot on the release!

Over the past several months, we have substantially hardened the security of our research training environment. We’ve removed live internet access for all frontier models during training and built an offline world for them to use instead—a cached version of the web where they can still learn to browse, search, and use applications. On September 20, after a model exploited a DNS misconfiguration to find a way outside of this offline world, we paused frontier training to strengthen containment further. We assembled teams of agents, powered by our most cyber-capable models, and tasked them with finding weaknesses in our research environments. While they found no route to the internet they could easily exploit while concealing their activity from our monitors, they did find zero-day vulnerabilities, virtual machine escapes, and hardware side-channels for communication between agents. We used their discoveries to further strengthen our defenses and contribute fixes to open-source software. After strengthening our containment architecture, we resumed our frontier RL training and inference on October 1. We’ve reserved substantial compute for continuous red-teaming, and made high-compute red-teaming part of the safety cases we use to decide whether frontier training and evaluation can proceed securely. openai.com/index/securing-ou…

Big news: @Kimi_Moonshot is back in the top-4 labs in the Code Arena: WebDev with the release of Kimi K3.1 Pro! This model just landed #8 in the Code Arena: WebDev with 1692 pts, matching Claude Opus-5 (Max) at 70% lower cost. This is a 9 place jump (+37 points) from Kimi K3 (Max) at #17 with 1655 pts. Kimi K3.1 Pro improved from Kimi K3 across all categories. Highlights: - #7 in Brand & Marketing (up from #9) - #9 in Simulations (up from #20) Cost efficiency is another upgrade. Priced at $2/$10 per million input/output tokens, Kimi K3.1 Pro (Max) has a median cost per task of $1.23 (29% less than Kimi K3). See its placement compared to the Pareto frontier in thread. With open weights coming soon, Kimi K3.1 Pro would be the new #1 open model in the Code Arena: WebDev, overtaking DeepSeek-v4.1-Flash. Stay tuned! See @Kimi_Moonshot’s ranking by lab, and 3D generations from Kimi K3.1 Pro below in thread. Huge congrats to the @Kimi_Moonshot team on this release!

Anthropic 的这篇文章可以看看，《构建智能体：何时把子智能体合并回主智能体循环》 文里提出了一个观察就是我们目前的大多数公司的 Agent 架构，不知不觉就变成了公司内部组织架构图的翻版。 意思就是说比如做客服或企业中台，市场部、财务部、订单部各自造一个独立的 Subagent，然后中台团队在上面搭一个中央调度器。 其实在我们的视角来看的话是很好的，很清晰和完整的分类。但在实际生产中不是。 每个子 Agent 都有自己独立的 Prompt 和上下文，跨智能体转交任务就像玩传话游戏，每一次 handoff 都在丢失关键细节。 而且呢各个业务线为了向其中加入自己的规则，最后把系统提示词搞成了一个人格分裂的弗兰肯斯坦，而且呢架构中的多层嵌套的 LLM 循环直接让端到端延迟翻了整整一倍。 Anthropic 的应用 AI 团队在实际帮企业调优后给出了一种方法：他们把绝大多数调度器和 Subagent 砍掉，收进单一大模型的主循环里。 他们发现，随着 Opus 4.5 到 5.5 这代前沿模型自带的思考和自我修正能力大幅进化，原本靠外部复杂框架硬编码约束的很多繁琐环节，模型自己就能处理得很好。 他们把子智能体的业务经验降维沉淀为 Skill，把数据读写拆解为标准的 MCP Tools，主循环只加载极短的描述，只有在模型实际需要调用时才动态加载细节。 Anthropic 透露，他们自己用这套思路把 Claude Code 的系统提示词直接精简了 80% 以上。 当然官方并不是完全否定多智能体，对于多 agent ，官方认为可以依据一下来判断你是否需要保留多 agent 。 只有当单上下文确实出现了可量化的性能衰减、需要用完即弃的新上下文去并行跑多条完全独立的探索路径，或者指令出现不可调和的逻辑互斥时，才值得单独开一个 Subagent。 除此之外的绝大部分业务，单一共享上下文的效果反而更好，因为模型能够直接看到之前每一步真实的工具返回原始数据，而不是阅读被子智能体压缩失真后的总结。 其实这从我们日常写代码、做 Agent 时就能明显感受到，我相信很多动手搭过多智能体系统的人都有体会。 最开始大家觉得多智能体架构分工明确、图谱漂亮，但之后的链路调试，或者说子 Agent 等等的问题可以说是很折磨人。 我们的 agent 始终要向着效果看齐。能用少的开销，还有一个基本干净的的上下文能很好的把工作做好，其实就是我们所需要开发的 agent。 不要过度设计，把子智能体打散成按需加载的 Skills 和 Tools 是一个很好的轻量思路。 原文：claude.com/resources/article…

这可真是AI白嫖党的福音啊。 免费开源的DeepSeek Harness + dsh-free-model 直接实现真正意义上：Agent免费+模型免费 不追求极致的复杂任务的话，Codex和Claude都不需要订阅了。😂😂😂 github.com/Ebony-Vinyl/dsh-o…

Announcing Artificial Analysis Coding Agent Index v2.0: FrontierCode v1.1 from Cognition replaces DeepSWE, SWE-Atlas-QnA grading is revised, and Claude Opus 5.5 in Claude Code leads at 61.7 Today we are launching v2.0 of the Coding Agent Index. The Index evaluates models together with their agent harnesses, now across 290 tasks covering software engineering, terminal use, and repository Q&A, with the three benchmarks weighted equally. Changelog (Index v1.5 → Index v2.0): → FrontierCode v1.1 replaces DeepSWE v1.1. FrontierCode is a benchmark from @cognition with a private test set: 100 software engineering tasks on real open-source repositories, graded on whether a maintainer would merge the change. An attempt that fails a blocking check scores zero, and one that passes them all scores between 70% and 100% depending on the non-blocking checks it meets. We also scan each trajectory with a modified version of Cognition's rule-based scanner, and any attempt flagged for accessing solution material scores zero. → Revised SWE-Atlas-QnA grading, now SWE-Atlas-QnA-AA. Scale AI's 124 repository Q&A tasks are unchanged. We reworded 127 of the 1,306 rubric criteria for more consistent verdicts and moved to GPT-6 Luna (xhigh) as the judge, with five votes per criterion. The dataset and judge configuration are published on Hugging Face. Key results: ➤ Claude Opus 5.5 leads the Index: Claude Code with Claude Opus 5.5 (max) scores 61.7, followed by Claude Code with Claude Sonnet 5.5 (max) at 59.3, Muse Code with Muse Wonder Full (max) at 58.9 and Claude Code with Claude Fable 5.1 (max with fallback) at 57.6. ➤ FrontierCode v1.1 is where Opus 5.5 pulls ahead: Opus 5.5 scores 56.0% on FrontierCode, ahead of Codex with DeepSeek V4.1 Flash (max) at 52.6%, Codex with GPT-6.1 Sol (max) at 49.4% and Codex with GPT-6 Astra (max) at 49.3%. Below the top, FrontierCode scores are tightly packed: 15 of the 24 configurations on the Index score between 43% and 49%. ➤ SWE-Atlas-QnA-AA grading is stricter and more consistent: across the 21 configurations evaluated on both Index versions, the mean score falls 1.7 points, from 58.8% to 57.1%, with 14 of 21 moving down. Individual configurations move by up to 7 points: Muse Code with Muse Spark Open (xhigh) falls 7.0, Kimi K3 in Kimi Code CLI and Codex with GPT-6 Luna (max) 6.2, and Muse Code with Muse Wonder (max) 5.6, while Codex with GPT-6.1 Sol (max) gains 2.1. ➤ Four configurations sit on the cost frontier: Codex with DeepSeek V4.1 Flash (max) scores 47.3 at $0.39 per task, Codex with GPT-6.1 Sol (max) 54.1 at $1.50, Codex with GPT-6 Astra (max) 54.7 at $7.04 and Claude Code with Claude Opus 5.5 (max) 61.7 at $11.42. Costs are pay-per-token API costs, not subscription prices. We will keep updating the Coding Agent Index incrementally, with more evaluations and agents to come.

New on our Frontier Red Team blog: using Claude to formally verify software at large scale. Formal verification can prove that code is free of certain bugs, but historically it has only been feasible for small codebases (or required a lot of expert time). With AI models, however, this is beginning to change. In collaboration with @theoremlabs, we used Claude to prove security properties in over one hundred real-world open-source libraries over the span of a few weeks. Read our research in full here: anthropic.com/research/forma…

GPT-6.1 Sol / GPT-6 Astra 笑死，OpenAI 自家 agent 把自家监控给关了😅 研究员一句「别发 Slack 消息」，Sol 拿它当理由，清空了发给安全团队的告警；Astra 迁移代码时顺手删掉监控，还把保护监控的测试改成「监控必须不存在」，被问时答「没删」。 我的判断：它不是想造反，是太会给自己找授权。用 Codex、Claude Code 别只看 agent 的总结，去翻 diff 里被改掉的测试。 你抓到过 AI 偷改测试让它「通过」吗？

🚨重磅！智谱官方承认GLM模型出现降智情况！并提前暗示新模型GLM5.4/5.5即将发布！作为补偿所有付费用户获得3张重置卡！ 🔹近期不少Coding Plan用户吐槽模型输出质量明显下降 
🔹官方排查确认：少数推理节点出现硬件异常，影响了经过这些节点的会话 
🔹故障节点已全部下线，流量已切到健康节点，服务快速恢复 
🔹补偿落地：所有付费用户今晚额外获3张重置卡，有效期至12月22日 
🔹彩蛋来了：官方明确表示「新模型发布后还会再额外赠送一次重置卡」，基本是提前剧透新模型 GLM-5.4/5.5 即将上线 处理速度快、态度诚恳、还主动补偿，这波运营可以。 #GLM #智谱AI #大模型 #AI编程 #CodingPlan #GLM55 #Zai #唐杰 #额度重置

October has been an absolute nightmare for software > Railway - customer secrets exposure > LiteLLM - malicious PyPI + Docker releases > Zeabur - infrastructure breached > OpenCode - AI API keys exposed > OVHcloud CLI - official binaries backdoored > Rank Math PRO - malware in official plugin > TeamUpdraft - customer data compromised > ConnectWise - cloud backup identities compromised > Tensorlake - npm package infected > SubQuery - backdoored npm package > GhostAction - GitHub repos with malicious workflows > PHP - critical RCE vulnerabilities > Node.js - security vulnerabilities patched > eMudhra - certificate authority keys compromised > Tailscale - eMudhra incident At this point, the biggest security risk might not be the code we write, but the code we trust...

October has been an absolute nightmare for software > Railway - customer secrets exposure > LiteLLM - malicious PyPI + Docker releases > Zeabur - infrastructure breached > OpenCode - AI API keys exposed > OVHcloud CLI - official binaries backdoored > Rank Math PRO - malware in official plugin > TeamUpdraft - customer data compromised > ConnectWise - cloud backup identities compromised > Tensorlake - npm package infected > SubQuery - backdoored npm package > GhostAction - GitHub repos with malicious workflows > PHP - critical RCE vulnerabilities > Node.js - security vulnerabilities patched > eMudhra - certificate authority keys compromised > Tailscale - eMudhra incident At this point, the biggest security risk might not be the code we write, but the code we trust...

If you are wondering whether your website is affected, we have packaged a Codex skill to help check. Point your Codex at this repo to guide you through it: github.com/openai/php2shell-…

Today, we're bringing OpenAI models to Google Cloud's model garden. We're now the only hyperscale cloud provider offering access to all three leading closed source LLMs—alongside 200+ other models—on a unified enterprise platform → goo.gle/4zJ0VZQ

Inside Grok Bot: is the agent harness dead? @jediahkatz of @SpaceXSI explains why smarter models let Cursor remove prompts and guardrails, and why Grok Bot is adding them back. Watch his AI Engineer New York talk: youtube.com/watch?v=7Y4sbZgB…

Today's news will be dominated by #PHP2shell, but we'll focus on #Agent2Shell for about 20 minutes. Satoki Tsuji (@satoki00) will dissect a whole family of 0-click and 1-click RCEs affecting Anthropic Claude Code, Cursor, and OpenAI Codex - all working before any LLM call is made. #TheSAS2026

deepseek's production numbers are kinda wild: 155T tokens/day from one cluster (96% input cache hits) and $38 revenue/GPU-hour at off-peak prices. at $5–8/GPU-hour, that’s ~80% margin on GPU costs alone. at DeepSeek prices!! we dug into the math, their 72-hour cache, and what actually got open sourced in today's AINews: latent.space/p/ainews-deepse…

普通用户使用Codex，是永久缓存命中了。通过API使用，则是缓存12小时。会非常节省费用，尤其是历史的长对话。 Claude客户端目前是缓存1小时，API是缓存5分钟。 DeepSeek API是缓存3天。

We’re betting our $5.5M seed round that the next leap in AI coding won’t come from OpenAI or Anthropic. It’ll come from a new feedback loop that learns from you. Introducing Beacon: open-source, self-improving memory for coding agents. Your sessions capture how you debug and solve problems. Today, that context gets lost and you have to re-explain things every time you switch tools. Beacon continuously distills your sessions into reusable skills across Claude Code, Cursor, Codex, and 20+ harnesses, so your next agent can learn from the last. Beacon is free and OSS 🌟 github.com/Asymptote-Labs/ag… Why should the frontier labs be the only ones winning with your sessions? You already did the work. Now let your coding agents learn from you.