Latest

Every published item, in order · 78 items

Nov 10

Chinese contentAnthropic 发布 Claude Fable 5.2，登顶 Artificial Analysis 智能指数

Anthropic 的新旗舰模型，科学推理和编程基准刷了一堆纪录，比如 HLE-Diamond 拿到 80%。有意思的是调低 effort 档反而更划算，输出 token 用得更少还能赢 Opus 5.5。做科学计算、代码开发的团队值得看下这笔成本账。

#Claude Fable 5.2 #Anthropic #Gemini 4 Carbon #推理模型

Official accountArtificial Analysis@ArtificialAnlysSource

Nov 9

Papers US vs China02:48

Chinese content清华与 Moonshot AI 提出 TCSExam 基准：GPT-6 Astra 定理证明率达 70.0%

清华和 Moonshot AI 搞了个 TCSExam 测试集，GPT-6 Astra 证定理能拿 70.0%，但 30 个开放猜想全挂，差距很直观。

#TCSExam #GPT-6 Astra #Kimi K3.1 Pro #Moonshot AI

rohanpaul_ai@rohanpaul_aiSource

Nov 8

Industry BuildersOfficial09:52

Chinese content华为昇腾在128卡Atlas 950 SuperPoD上完成Kimi K3.1 Pro LoRA微调验证

昇腾把月之暗面的2.8万亿参数Kimi K3.1 Pro跑通了LoRA微调，128卡集群加Swap Core缓存提速15%，国产算力训练国产开源模型又添一例。

#Kimi K3.1 Pro #华为昇腾 #Atlas 950 #LoRA

Primary sourcepandaily@[email protected] (Pandaily)Source

Models US vs China03:02

Chinese contentKimi K3.1 Pro 开源，百度 ERNIE 5.5 Flash、腾讯 Hy4 相继开放

Kimi K3.1 Pro 能直接下载了，百度和腾讯也开放了自家模型，不过分数都是官方自己报的，得留个心眼。

#Kimi K3.1 Pro #ERNIE 5.5 Flash #腾讯 #开源模型

GenkSource

Nov 7

Models US vs China07:56

Chinese contentArena 周报：Kimi K3.1 Pro、腾讯 Hy4、Gemini Omni 1.2 Flash 登顶多项榜单

LMArena 这周榜单很热闹：Kimi K3.1 Pro 开源拿下 Agent 第一，腾讯 Hy4 成本才 $0.57，Gemini 视频、OpenAI 图像都有新作，一条看全。

#Kimi K3.1 Pro #Hunyuan #Gemini #GPT Image

lmarena.ai@lmarena_aiSource

Models BuildersSelected01:14

VIALS-AA leaderboard ranks models on life science visual reasoning

The 29× cost gap between frontier models with a 3-point score spread changes the economics of choosing a model for scientific agent workloads. Teams building lab-automation or biotech analysis tools can use the per-task cost and turn counts to pick models that fit their budgets. The 19-point deficit of open weights models like Kimi K3.1 Pro also marks a concrete ceiling for on-premise deployments.

#VIALS #Gemini 4 Carbon #GPT-6 #Claude Opus 5.5

Official accountArtificial Analysis@ArtificialAnlysSource

Models Builders71°00:02

Chinese contentKimi K3.1 Pro 获 TPU v7x day-0 支持

Kimi K3.1 Pro 发布当天就能在 Google TPU v7x 上跑了，用 vllm-torchtpu 即可部署，不依赖 NVIDIA GPU 的朋友可以看看。

#Kimi #vLLM #TPU #开源模型

Official accountvLLM@vllm_projectSource

Nov 6

Models Builders22:44

Chinese content摩尔线程完成 Kimi K3.1 Pro Day-0 适配，支持文本与图片推理

摩尔线程用自家 MTT S5000 卡跑通了 3 万亿参数的 Kimi K3.1 Pro，发布当天就完成适配，国产芯片推理链路已经打通了文本和图片。

#Kimi K3.1 Pro #摩尔线程 #MTT S5000 #月之暗面

IT之家Source

Models US vs ChinaSelected17:58

Tencent Hunyuan's Hy4 tops APEX-Agents open-weights leaderboard

The gap between Hy4 and the next open-weights model is nearly 8 points on Pass @1, which resets assumptions about where free-weights agent performance stands relative to closed models. Teams selecting models for professional-work agents—consulting, finance, legal workflows—can use the Mercor APEX-Agents leaderboard to shortlist candidates by domain rather than headline scores. Comparisons against Kimi K3.1 Pro and Qwen3.8-Max make it a practical benchmark for China open-weights options.

#Hy4 #Tencent Hunyuan #APEX-Agents #DeepSeek-V4.1-Flash-1017

Event

Hunyuan@TXhunyuanSource

Models BuildersSelected71°11:18

Tencent Hunyuan Hy4 Official Release Adds Vision and Computer Use

The upgrade comes from post-training and perception rather than scale, which means the same 770B base now handles screenshots, charts and full desktop workflows — a shift that matters more for agent builders than raw leaderboard chasers. Missing third-party evaluations mean teams should run their own tests before switching, and the free quota until 15 Nov 2026 makes a hands-on comparison against DeepSeek or Qwen low-cost to attempt.

#Hunyuan #Tencent #open-source models #agents

掘金本周最热@码事漫谈Source

Models US vs China74°08:42

Chinese content腾讯混元 Hy4 登顶 Agent Arena 中文模型榜首

腾讯混元 Hy4 冲到 Agent Arena 中文第一，单任务成本比 Kimi K3.1 Pro 便宜六成多，分数还几乎持平，跑智能体任务可以试试它。

#Hunyuan #Kimi K3.1 Pro #Agent Arena #智能体

lmarena.ai@lmarena_aiSource

Products Builders06:48

Chinese contentKimi K3.1 Pro 上线 Devin Desktop 与 Devin CLI

月之暗面的 Kimi K3.1 Pro 进了 Devin 的桌面端和命令行，编程基准分数比上一代还高了一点，写代码的可以试试。

#Kimi #Devin #FrontierCode 2.0 #编程助手

Windsurf@windsurf_aiSource

Models US vs China01:38

Chinese contentScience's Last Exam 基准发布：91 项任务测端到端科研能力

新基准 SLE 用 91 个真实科研任务测模型，Claude Opus 5 以 67 分领先 GPT-6.1 Sol 和 Kimi K3.1 Pro，做科研自动化评估的可以看看。

#Science's Last Exam #Claude Opus 5 #GPT-6.1 Sol #Kimi K3.1 Pro

Aran Komatsuzaki (论文推介)@arankomatsuzakiSource

Products Builders00:51

Chinese content11月5日AI快讯：OpenAI发布Images 2.6，Kimi K3.1 Pro开放权重

一天的新闻打包好了：Kimi K3.1 Pro能下载了，OpenAI的模型一口气修了948个Linux漏洞，做开发的可以看看Codex和Xcode这几条。

#OpenAI #Kimi K3.1 Pro #Gemini API #GitHub Copilot

TestingCatalog@testingcatalogSource

Models US vs ChinaSelected72°00:48

Tencent Hunyuan Hy4 scores 35 on Artificial Analysis Index v5

The striking detail is the cost-efficiency: at US$0.78 per Intelligence Index Task, Hy4 delivers Grok 4.7-level intelligence at a tenth of the price, which shifts the calculus for teams running high-volume agentic workloads. Developers building coding agents should weigh it against Claude Haiku 5.5, which scores higher but costs nearly twice as much; the 89% hallucination rate makes it a weaker pick for knowledge-retrieval tasks. API availability and USD pricing put it in direct reach of Southeast Asian startups comparing China-sourced models.

#Tencent Hunyuan #Hy4 #Artificial Analysis #benchmarks

Official accountArtificial Analysis@ArtificialAnlysSource

Models US vs China00:26

Chinese content腾讯发布 Hy4 多模态模型，可自主操作软件

腾讯出了个能看图还能自己操作软件的 Hy4，中国模型里排第二，不过离最强模型还有距离。

#Hy4 #Tencent #Kimi K3.1 Pro #多模态

Official accountDecoder@Jonathan KemperSource

Models Builders79°00:13

Chinese contentKimi K3.1 Pro 开放权重上线 SiliconFlow

月之暗面的 Kimi K3.1 Pro 开放权重上线 SiliconFlow 了，跑分挺猛，做编程智能体和 3D 建模的可以试试。

#Kimi #Moonshot #SiliconFlow #开源模型

Official accountSiliconFlowAI@siliconflowaiSource

Nov 5

Papers Builders23:42

Chinese content协作智力基准发布：18 个模型当队友，最高只拿 82.2 分

斯坦福他们测了 18 个模型当队友的水平，最强的 Gemini 4 Argon 也才 82.2 分，冲突一开会就妥协，挺扎心的。

#协作智力基准 #Gemini 4 Argon #Claude Opus 5.5 #Kimi K3.1 Pro

shao__meng@shao__mengSource

Models US vs ChinaSelected18:56

Tencent Hunyuan Hy4 scores 35 on Intelligence Index v5

Hy4 achieves parity with Grok 4.7 at one-tenth the cost, with a Cost per Intelligence Index Task of US$0.78 versus far pricier Western peers. The main caveat is an 89% hallucination rate on AA-Omniscience v1.1 against Kimi K3.1 Pro's 37%, so agentic coding is the safer use case than knowledge retrieval. Teams benchmarking China-based models against Claude, Grok or Qwen options should weigh that trade-off directly.

#Tencent Hunyuan #Hy4 #Artificial Analysis #reasoning models

Hunyuan@TXhunyuanSource

Products BuildersOfficial15:13

Chinese content阿里云灵骏真武 M890 SuperPOD 首日适配 Kimi K3.1 Pro

阿里云的灵骏 M890 SuperPOD 在 Kimi K3.1 Pro 发布当天就能跑这个 2.8T 参数的模型，用 vLLM 和 SGLang 部署，要自建推理集群的可以看看这套方案。

#Kimi K3.1 Pro #阿里云 #Moonshot AI #vLLM

Primary sourcepandaily@[email protected] (Pandaily)Source

Models Builders14:35

Chinese contentOpenAI 参与 Linux 内核安全，Kimi K3.1 Pro 开放权重

这期日报一次讲三件事：OpenAI 做了 Linux 内核安全，Kimi K3.1 Pro 放出了开放权重，还有 Ricursive 的 Ripple-65B，想追开源模型的可以看看。

#Kimi K3.1 Pro #Ripple-65B #OpenAI #开源模型

Official accountLatent Space (swyx)@swyx .aiSource

Industry US vs ChinaOfficial10:29

Chinese contentHuawei Ascend 为 Kimi K3.1 Pro 提供首发日支持

Kimi K3.1 Pro 这个 2.8 万亿参数模型发布当天就能在 Huawei Atlas 和 Ascend 950 SuperPoD 上跑，vLLM 和 SGLang 都适配了。

#Kimi K3.1 Pro #Huawei #Ascend #vLLM

Primary sourcepandaily@[email protected] (Pandaily)Source

Models Builders79°08:50

Chinese content华为昇腾 0 Day 适配 Kimi K3.1 Pro 推理部署

Kimi K3.1 Pro 发布当天昇腾就能跑推理了，2.8 万亿参数、百万上下文，用 vLLM Ascend 或 SGLang 可直接部署。

#Kimi K3.1 Pro #昇腾 #华为 #月之暗面

IT之家Source

Models US vs China84°08:04

Chinese content月之暗面开放 2.8 万亿参数 Kimi K3.1 Pro 权重，登顶 Agent Arena 开放模型榜

月之暗面兑现承诺放出了 Kimi K3.1 Pro 权重，智能体任务实测比 K3 强了一截，编程成绩已经贴着 Claude Opus 5.5 打，做 Agent 的可以关注。

#Kimi #Moonshot AI #开源模型 #智能体

IT之家Source

Models US vs China05:32

Chinese contentMoonshot 开放 Kimi K3.1 Pro 权重，登顶 Agent Arena 开源模型榜首

月之暗面把 K3.1 Pro 权重放出来了，Agent Arena 上开源第一，Shell 命令成功率和听指挥能力都涨了不少，玩开源智能体的可以试试。

#Kimi K3.1 Pro #Moonshot AI #Agent Arena #开源模型

lmarena.ai@lmarena_aiSource

Models US vs ChinaSelected04:42

Kimi K3.1 Pro lands on Databricks at launch

Day-0 availability on Databricks signals Moonshot pushing Kimi into enterprise data platforms rather than consumer chat apps, putting it next to 70+ alternatives where procurement teams actually evaluate models. Unity Gateway means companies can A/B Kimi against rivals without code changes — useful for cost benchmarking. Enterprises with Lakehouse data estates, including those in Singapore and ASEAN running Databricks, can try it immediately.

#Kimi K3.1 Pro #Databricks #Moonshot (Kimi)#agents

Official accountDatabricks@databricksSource

Models Builders77°04:39

Chinese contentKimi K3.1 Pro 权重上线 HuggingFace，3T 级参数开源

月之暗面把 Kimi K3.1 Pro 权重放到 HuggingFace 了，3T 参数的多模态智能体模型，能自己下载部署。

#Kimi K3.1 Pro #Moonshot #HuggingFace #开源模型

TestingCatalog@testingcatalogSource

Models US vs China83°03:21

Chinese contentMoonshot AI 发布 2.8T 参数 Kimi K3.1 Pro 权重

月之暗面把 2.8T 参数的 Kimi K3.1 Pro 权重直接开源了，4-bit 发布省存储，编程基准分数贴近 Claude Opus 5.5，纯知识题却差 GPT-6 一截，许可证对云托管商还有限制。

#Kimi #Moonshot AI #Terminal-Bench #开源模型

rohanpaul_ai@rohanpaul_aiSource

Models BuildersSelected02:23

Agent Arena data shows 66% cost gap in hour-long agent tasks

For hour-long agent workloads, the cheapest model is not automatically the right one: Claude Opus 5.5 scores higher but costs twice as much per hour as Gemini 4 Carbon. Teams running long-horizon agents can use these per-hour cost and time figures to match model choice to budget and quality needs. The 55-minute spread among frontier models also suggests task duration, not just accuracy, should enter model selection criteria.

#Claude Opus 5.5 #Gemini 4 Carbon #GPT-6 Astra #Kimi K3.1 Pro

lmarena.ai@lmarena_aiSource

Models US vs China02:23

Chinese contentAgent Arena 排行：复杂任务上 Gemini 4 Argon 与小米 MiMo 表现突出

Agent Arena 拆了成本维度看模型：贵任务上 OpenAI 和 Anthropic 不再通吃，Gemini 4 Argon 和小米 MiMo 顶上，选型前值得对号入座。

#Agent Arena #Gemini 4 #MiMo-V2.6 #Claude Opus 5.5

lmarena.ai@lmarena_aiSource

Models US vs China02:22

Chinese contentAgent Arena 数据：头部模型长任务耗时差距达 55 分钟

Agent Arena 拿 5 家实验室的旗舰模型跑了长任务对比，Claude Opus 5.5 能连续干近 2 小时，Kimi 最快但也最短，差距挺直观。

#Agent Arena #Claude Opus 5.5 #Kimi K3.1 Pro #GPT-6 Astra

lmarena.ai@lmarena_aiSource

Models US vs ChinaSelected01:22

Kimi K3.1 Pro rolls out on Ollama Cloud with zero data retention

The zero-data-retention and US-hosting terms matter for teams that couldn't previously route sensitive workloads to a Chinese open-weight model. Developers can now self-host or run K3.1 Pro through Ollama Cloud as an alternative to renting closed APIs from OpenAI or Anthropic. Weight availability on Hugging Face also means inference providers beyond Ollama can deploy it.

#Kimi K3.1 Pro #Moonshot #Ollama #open-source models

ollama@ollamaSource

Models US vs China01:13

Chinese contentMoonshot AI 发布 Kimi K3.1 Pro，Together AI 提供托管服务

Moonshot AI 出了 Kimi K3.1 Pro，Together AI 上就能直接调用， podcast 里有他们自己人讲细节，想用的话可以先听听。

#Kimi K3.1 Pro #Moonshot AI #Together AI #开源模型

Official accountTogether AI@togethercomputeSource

Models US vs China00:30

Chinese contentMoonshot AI 开源 Kimi K3.1 Pro 权重，本地部署需 8 张 B300

Kimi K3.1 Pro 权重开放下载了，2.8 万亿参数，不过想自己跑得准备 8 张 B300，约 60 万美元起步。

#Kimi #Moonshot AI #开源模型 #Hugging Face

kimmonismus@kimmonismusSource

Models US vs China76°00:19

Chinese contentKimi K3.1 Pro 上线 Together AI 平台

月之暗面把最强的 Kimi K3.1 Pro 放上了 Together AI，100 万上下文还能自己微调，国内模型出海又多一个入口。

#Kimi #Together AI #Moonshot AI #多模态

Official accountTogether AI@togethercomputeSource

Models Builders00:13

Chinese contentKimi K3.1 Pro 登陆 vLLM，支持 1M token 上下文

Moonshot 新的 Kimi K3.1 Pro 发布当天就能用 vLLM 自己部署了，1M 上下文，Terminal-Bench 4.0 跑 65.7%，自己搭推理服务的话可以试试。

#Kimi #K3.1 Pro #vLLM #开源模型

Official accountvLLM@vllm_projectSource

Models Builders00:07

Chinese contentKimi K3.1 Pro 在 Databricks 上线，Moonshot AI 与 Databricks 成发布合作伙伴

Databricks 第一时间上线了 Moonshot 的 Kimi K3.1 Pro，自家团队已经拿它写代码了，开源模型又多一个能打的选项。

#Kimi K3.1 Pro #Moonshot AI #Databricks #开源模型

Yuchen Jin@Yuchenj_UWSource

Models Builders71°00:06

Chinese content月之暗面开源 Kimi K3.1 Pro 模型权重

Moonshot 把 Kimi K3.1 Pro 的权重放出来了，想本地跑或者自己微调的可以去 Hugging Face 下载试试。

#Kimi #Moonshot #开源模型

Gorden Sun@Gorden_SunSource

Nov 4

Models US vs China76°21:16

Chinese content腾讯混元发布 Hy4：新增视觉与 Computer Use，WorkBuddy 限时免费

腾讯混元的 Hy4 正式版能看图能操作电脑了，还练过 Blender 和 CAD，这两天在 WorkBuddy 里限免，可以白嫖试试跟 Kimi 的差距。

#Hunyuan #Hy4 #腾讯 #Computer Use

shao__meng@shao__mengSource

Models US vs ChinaSelected16:09

Tencent Hunyuan Hy4 Ranks #14 on Agent Arena

Hy4 matches a top open model on agentic performance at roughly a third of the cost per task, shifting the price-performance frontier for long-horizon agent workloads. Teams building finance, frontend, research or office automation agents can try it now via Tencent's preview page and should benchmark it against Kimi K3.1 Pro before committing. If Tencent open-weights the model as signalled, it becomes a strong free-tier candidate for cost-sensitive deployments across ASEAN and beyond.

#Hunyuan #Hy4 #Kimi K3.1 Pro #agents

lmarena.ai@lmarena_aiSource