Models11月11日 20:10

腾讯 Hunyuan Hy4 上架 OpenRouter

Hunyuan@TXhunyuan

Post· 2 sources

Why we picked it

腾讯 Hunyuan 的 Hy4 在 OpenRouter 上能直接调了，Terminal Bench 4.0 跑到 44.4 分。搞自动化智能体的可以拿它跟同类模型跑一遍对比，看 OSWorld 2.1 的 78.9 分在自家任务里什么水平。

腾讯 Hunyuan 的 Hy4 模型现已通过 OpenRouter 提供服务。官方公布的三项基准成绩为：Terminal Bench 4.

Body

0 得 44.4 分，OSWorld 2.

1 得 78.9 分，AutomationBench v1.

0.6 得 56.

2 分。这些分数集中在终端操作与计算机自动化任务上，说明该模型的定位偏向智能体执行场景，而非单纯的对话或文本生成。

Hy4 的详细技术说明发布在 hy.tencent.

ai/research/hy4。通过 OpenRouter 分发意味着开发者可以在统一 API 下调用该模型，与其他厂商的模型做横向对比或做路由切换。

Original · Hunyuan

Hy4 on OpenRouter Models on OpenRouter @ModelsOnOR Tencent's numbers: 44.4 on Terminal Bench 4.0, 78.9 on OSWorld 2.1 and 56.2 on AutomationBench v1.0.6. See more: hy.tencent.ai/research/hy4 🔗 View Quoted Tweet📊 6 ⚡

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Models" desk, focusing on China's AI industry and domestic foundation model ecosystem. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include Hunyuan, Hy4, OpenRouter, 智能体.

Source attribution: curated from Hunyuan (@TXhunyuan). Cross-source corroboration: 2 independent sources have covered this development (including Decoder), confirming industry-wide signal strength.

Decoder11-09 15:44Source

#Hunyuan #Hy4 #OpenRouter #智能体 #腾讯

Keep reading

Recent stories with the same tag or section

Industry BuildersMulti-source23:58

Chinese content腾讯 AI 负责人姚顺雨：AI 生化武器风险临近，但封闭模型不是答案

腾讯姚顺雨提了个有意思的中间路线：与其学 Claude 那样暴力拦截，不如像 Hy4 那样在训练数据里直接删掉危险知识。开源模型安全该怎么做，这篇文章给了个具体方案，做安全和对齐的值得看看。

#Tencent #Claude #AI安全 #开源模型

Event

Official accountDecoder@Maximilian SchreinerSource

Models US vs China18:03

Chinese content腾讯混元 Hy4 开源实测：搭配 WorkBuddy，多模态 Agent 表现亮眼

腾讯把混元 Hy4 开源了，我实测配 WorkBuddy 跑多模态 Agent 任务，交付效果相当能打，感兴趣可以看看实测细节。

#Hunyuan #WorkBuddy #开源模型 #多模态

Event

量子位@闻乐Source

Models US vs ChinaMulti-source07:56

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

Event

lmarena.ai@lmarena_aiSource

Models US vs ChinaMulti-sourceSelected07:54

Tencent Hunyuan Hy4 tops Agent Arena as best Chinese model

Cost-per-task of US$0.57 with top-5 agentic performance makes Hy4 a strong price-performance option for agent workloads. Developers evaluating Chinese alternatives to US frontier models can compare it directly on Agent Arena's session-based leaderboard. The model is relevant to ASEAN teams building on Tencent's ecosystem, where Hunyuan integration is already common.

#Tencent Hunyuan #Hy4 #Agent Arena #agents

Event

lmarena.ai@lmarena_aiSource

Models US vs ChinaMulti-source20:58

Chinese contentOpenDesign 设计基准：腾讯 Hunyuan Hy4 排名第 4，仅落后 GPT-6.1 Sol 0.1 分

腾讯 Hunyuan 的 Hy4 在 OpenDesign 设计基准排第 4，只差 GPT-6.1 Sol 0.1 分，成本还低 41%，做设计的可以看下榜单。

#Hunyuan #Hy4 #GPT-6.1 #设计基准

Event

Hunyuan@TXhunyuanSource

Models US vs ChinaMulti-sourceSelected17:58

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