---
格式版本: 2
标题: "Introducing the All-New Qoder"
原文链接: "https://www.alibabacloud.com/blog/introducing-the-all-new-qoder_603515"
发布日期: "2026-08-31"
发布时间校准状态: "found"
发布时间需复核: "否"
发布时间来源: "rule:configured_publication_date_rule"
发布时间证据: "alibaba-cloud-news-publication-date html:original: Alibaba Cloud Community August 31, 2026"
发布时间校准原因: "信源发布日期识别规则直接确认发布时间"
发布时间校准置信度: "high"
发布时间候选数量: 2
发布时间严格候选数量: 2
发布时间原页读取状态: "source template page reused from URL open"
发布时间未找到原因: ""
发布时间校准时间: "2026-08-31T20:34:45+08:00"
发布时间仲裁状态: "skipped"
发布时间仲裁尝试次数: 0
发布时间仲裁耗时毫秒: 0
发现时间: "2026-08-31T20:31:16+08:00"
入库时间: "2026-08-31T12:35:27.173Z"
来源平台: "固定入口"
搜索渠道: "fixed_url"
搜索词: "https://www.alibabacloud.com/blog"
匹配关键词:
  - "performance"
  - "AI"
相关厂家:
  - "阿里"
相关专家:
  []
内容类型: "网页"
抓取工具: "Free Fetch + Defuddle"
清洗工具: "Defuddle Markdown + Defuddle/Readability 正文提取"
原始附件:
  []
AI摘要: "阿里云发布全新Qoder Agent Desktop，以任务为中心、跨终端与IDE协作，让开发者及非开发者通过自然语言即可完成原型、测试、运维等任务。内置多模型路由与70+专家套件、40+连接器，并支持长期记忆和自主执行，减少人工介入。"
AI摘要模型: "ali-deepseek-v4-flash"
AI摘要时间: "2026-08-31T23:37:21.197Z"
AI优质: "否"
AI打分: 34
AI分档: "非优质"
AI质检状态: "不通过"
AI打分理由: "正文主线是Qoder编码代理桌面应用及任务委派、多模型路由和连接器能力，并非超节点、AI Rack或机架级基础设施。来源为阿里云官方博客，对自有产品具有较高可信度；固定知识库未发现同产品记录，但未命中不能证明首次发布。新增事实仅包括桌面应用“今日上线”、70+专家套件、20000+技能和40+连接器等应用层信息，没有机架拓扑、互连、供电、液冷、RAS、量产客户或规模部署证据。命中主题范围不匹配及应用导向强否决，当前页面不适合作为超节点业务情报源。"
AI质检模型: "gpt-5.6-sol"
AI质检时间: "2026-09-01T07:40:21+08:00"
AI主题相关性: 0
AI来源权威性: 12
AI新颖性: 10
AI技术细节: 1
AI商业部署信号: 2
AI完整性: 9
AI评分提示词版本: "v17-精简生产版"
AI评分提示词SHA256: "48fb9777f386026761b4873eaff30807694fb11e9b352d7c69bf2dfde750cc7d"
AI评分知识库版本: "knowledge_base_v1-20260819+runtime.63"
AI评分知识库SHA256: "b2a4248d43868dee4c86d2b502d43668e260aea6b69d15f73a5dfe53287ca63f"
AI评分知识库检索词: "[\"阿里\",\"https://www.alibabacloud.com/blog\",\"Intel\",\"IDE\",\"SOTA\",\"APIs\"]"
AI评分知识库命中: "[{\"id\":\"runtime-5581f72bc66fda8e6c6bf711\",\"title\":\"Advancing Standards-Based AI Fabric RAS Through COSMOS Integration\",\"sourceType\":\"ai_excellent_article\",\"time\":\"2026-08-10\",\"matchedTerms\":[\"Intel\",\"IDE\",\"APIs\"],\"rank\":-7.40143553256192},{\"id\":\"july-correct-0123\",\"title\":\"ODCC分享 | UALink联盟Kurtis：开放Scale-Up互连加速构建可部署AI超节点\",\"sourceType\":\"labeled_article\",\"time\":\"2026-07\",\"matchedTerms\":[\"阿里\",\"Intel\"],\"rank\":-7.014673833652701},{\"id\":\"july-correct-0130\",\"title\":\"ODCC分享 | UALink联盟Kurtis：开放Scale-Up互连加速构建可部署AI超节点\",\"sourceType\":\"labeled_article\",\"time\":\"2026-07\",\"matchedTerms\":[\"阿里\",\"Intel\"],\"rank\":-7.014673833652701},{\"id\":\"july-correct-0081\",\"title\":\"AAI 2026: 6th Gen AMD EPYC Server CPUs Power the Agentic Data Center\",\"sourceType\":\"labeled_article\",\"time\":\"2026-07\",\"matchedTerms\":[\"Intel\",\"IDE\"],\"rank\":-5.477279841981439},{\"id\":\"historical-may-024\",\"title\":\"OpenAI、Microsoft等围绕MRC协议构建更大规模AI以太网训练网络\",\"sourceType\":\"curated_item\",\"time\":\"2026-05\",\"matchedTerms\":[\"Intel\"],\"rank\":-5.169105809474991}]"
采集批次: "2026年8月31日20点21分03秒"
采集批次ID: "20260831-202103-550"
去重键: "https://www.alibabacloud.com/blog/introducing-the-all-new-qoder_603515"
---

### Learn More about Qoder

### Explore Qoder for Enterprise

---

AI is reshaping the craft of software development. Three shifts are already underway.

## What We Work On: From Code to Tasks

Once, opening a tool meant facing files, directories, and functions. Now you hand over a task: turn this data into a report, turn this idea into a clickable prototype, trace last night's timeout.

Code is still there, but **it has become an agent's working material, not the thing you directly shape**. Tasks carry state and lifecycle. They run for hours, pause for your word, and keep going after you close the lid. **Centering on the task is the essential interface between people and agents.**

## How We Work: From Making to Delegating

When an agent can carry a task from start to finish, the human role changes. The scarce resource is no longer typing speed— **it is attention**. Five tasks in motion; which needs you now, which can wait, which is drifting off course?

Delegating and reviewing an agent's work calls for a different interface than writing code.

## Who Builds: From Programmers to Everyone

Code is the lingua franca of the digital world. **Coding is becoming a universal skill.**

Product managers need working prototypes. Designers need mockups that run. Testers need scripts from use cases. Operators need that weekly report automated. They know what they want and can judge the result—they shouldn't need to install a dev environment first. **So Qoder meets each role in the form they already know, so the shape of the product never blocks the use of agents.**

These three shifts point to one answer: a resident, task-centered desktop workbench where you can observe and step in. That is the Qoder Agent Desktop. It carries the same Qoder agent across terminal, IDE, and desktop. Where earlier tools asked, "Which files should we edit?" the workbench asks, "Is it done?"

## Why Qoder

### A Multi-Model Strategy That Picks the Right Mind for the Job

A new frontier model arrives almost every month. Qoder stays untethered to any single model, integrating the latest SOTA options as they appear.

Different tasks—quick fixes, deep refactoring, long-chain debugging, writing docs—demand different strengths. Experts can choose and configure. Everyone else can lean on **Ultimate, Performance, Efficient, and Auto** routing, which balances quality, speed, and cost.

**You choose when you care. Qoder chooses when you don't.**

### Harness: Turning Ideas Into Outcomes

Model intelligence only matters if it can be delivered. The same model in different harnesses can differ several-fold in whether it finishes the job. Real tasks are not single Q&As; they are long chains of decomposition, missteps, surprises, and recovery. The model thinks. **Harness sees it through.**

Qoder uses the latest agentic-loop technology to hold long tasks steady:

1. **A robust autonomous loop****: self-directed read-edit-verify-iterate, with cross-file changes and test-driven self-correction.**
2. **Clear guardrails****: tiered permissions and tool whitelists; sensitive actions require explicit approval.**
3. **Co-tuned models and engineering****: deeply adapted SOTA models, co-evolved with Qwen for long context, context compression, and precise tool use.**
4. **Continuous learning****: long-term memory and skill distillation across sessions.**

The result: longer, harder tasks with less hand-holding.

### Real Scenarios, Real Environments

Qoder is not only for programmers. Product managers turn ideas into interactive prototypes; designers make mockups live; operators automate weekly reports. Developers turn personal workflows into reusable apps.

We aim at real work, not demos. **Real tasks usually fail not from lack of ability, but from lack of context.** Qoder reads the actual workplace, so **you don't have to enumerate every assumption—the unclear parts, it investigates itself.** The more you use it, the deeper its model of your environment becomes. Model power is a commodity; context is your asset.

Through Browser Use and Computer Use, the agent acts in real environments, not just chat windows and codebases. Pause, take over, or point at output and ask for a fix. Qoder keeps the work visible and the choice yours.

### Ecosystem and Connectors: Reach More, Do More

Coding is the foundation, but the agent must reach into every place work happens. Qoder ships with **70+ expert suites, 20,000+ skills, and 40+ connectors**, and lets you turn your own workflows into reusable patterns. Connectors link code hosting, documents, data sources, collaboration tools, and live business systems—databases, internal APIs, tickets, monitoring. **An agent that cannot touch your real environment can only play with demos.**

One workbench now covers documents and proposals, data cleaning and analysis, design-to-code, ops automation, test scripts, and alert root-cause analysis. With Qoder, you do not learn to code; **you gain a team-sized capability that grows more personal over time.**

### Proactive: From You Driving to It Nudging

Instead of waiting for you to remember everything, Qoder brings the unhandled error, the missing test, the inconsistent logic to your attention. **You decide; it remembers.**

## This Is Just the Beginning

The Qoder desktop app launches today, but the road ahead is long.

Our aim is clear: **less human intervention, more trustworthy results.** The two must advance together—autonomy without trust breeds rework; trust without autonomy is merely a cautious assistant.

Next comes coordination across tasks, joint orchestration of people and agents, and turning a single success into a reusable capability.

**When coding becomes a universal skill, a coding agent becomes the power to turn ideas into reality.**

We want that power in everyone's hands.