Overview
Qoder is an Agentic AI programming platform launched by Alibaba, first released in August 2025, with version 1.0 officially launched on 2026/5/15, upgrading from an 'AI IDE' to a fully 'Autonomous Agent Development Workbench'.
The core concept is 'Autopilot Programming': developers only need to define the goal, and the entire process of execution, verification, and delivery is autonomously completed by the AI Agent. Qoder integrates the world's top programming models, provides ultra-long context capabilities (retrieving 100,000 code files at once), supports cross-project parallel tasks, and allows the creation of dedicated Agent teams. The concurrently launched QoderWork design workbench supports voice input to generate professional designs, achieving 'Design as Code'.
Major Update on 2026/6/8: Team Knowledge Engine, which upgrades AI from a simple Q&A tool to a collaborative partner that continuously accumulates team engineering experience and reuses knowledge across sessions and members. Note: Qoder and another Alibaba Cloud product line 'Qoder CN' (formerly Tongyi Lingma, renamed on 2026/5/20) are two related but independent brands—Qoder focuses on the global Agentic programming platform, while Qoder CN is positioned as a full-scenario agent matrix upgraded from domestic IDE plugins, both backed by Alibaba Tongyi Qwen3 series code models.
Latest Product Line Expansion: Qoder has evolved from a single programming platform into a complete product suite, including Qoder Desktop (autonomous development desktop), QoderWork (local-first, self-driven, secure AI work companion), QoderWake (24/7 AI employee), Qoder CLI (terminal-native AI programming partner), and Cloud Agents (fully managed cloud AI agent platform), covering all scenarios from programming to daily work.
Update on 2026/9/30: Jev Judgment Model Integration. Qoder integrated Jev, a System One judgment model from TypeSafe AI, into its Agent Harness: generative models handle understanding and planning while Jev makes bounded semantic judgments, cutting permission judgment latency in Auto Mode from 3.13 to 0.53 seconds and driving next-step selection in browser automation.
Key Features
- Team Knowledge Engine (New): Updated on 2026/6/8, AI continuously accumulates team engineering experience, reuses knowledge across sessions and members
- Jev Judgment Model Integration (New): Updated on 2026/9/30, Qoder integrated Jev, a System One judgment model from TypeSafe AI, into its Agent Harness: generative models handle understanding and planning while Jev makes bounded semantic judgments, cutting permission judgment latency in Auto Mode from 3.13 to 0.53 seconds and driving next-step selection in browser automation
- Quest Mode Independent Workbench: Upgraded to an independent window in version 1.0, integrating task management, status tracking, artifact tracing, and knowledge invocation; developers only need to define the goal
- Autonomous Agent Development: The entire process of execution, verification, and delivery is autonomously completed by the Agent, ushering AI programming into 'autopilot' mode
- Cross-Project Parallel Capability: Parallel scope expanded to cross-project and cross-repository, with multiple Workspaces running different Agent tasks simultaneously, providing a global view on one screen
- Ultra-Long Context: Retrieves 100,000 code files at once, offering the industry's strongest codebase context engineering capability
- Custom Expert Agent: Create dedicated Agent teams, configure domain knowledge, task skills, and external tool interfaces to fit business scenarios
- Spec-Driven Development: Requirements → Specifications → Automatic code conversion, automatically breaking down complex tasks into detailed design specifications and implementing them
- Deep Context View: View file directories, code changes, terminal output, and browser preview without leaving the current workflow
- Summary Delivery Checklist: Automatically generates a complete delivery summary after task completion, facilitating review and acceptance
- QoderWork Design Workbench: Supports voice input to generate professional designs, outputs runnable, editable, and deliverable artifacts on an infinite canvas, with 100+ built-in style references
- MCP Tool Ecosystem: Supports MCP (Model Context Protocol) extensions, allowing integration with a rich variety of external tools and services
- QoderWake 24/7 AI Employee: A secure, production-ready, continuously improving AI employee, providing unlimited virtual manpower for organizations
- Qoder CLI: A terminal-native AI programming partner, also serving as an extensible Agent engine
- Cloud Agents: A fully managed cloud AI agent platform, accelerating multi-scenario integration and achieving real business impact
- Qoder Desktop: An autonomous development desktop, built for real-world software construction
Use Cases
- Automated development where developers define the goal and the AI Agent autonomously completes execution, verification, and delivery
- Cross-project and cross-repository parallel development with multiple Workspaces running different Agent tasks simultaneously
- Team engineering knowledge accumulation and reuse across sessions and members via the Team Knowledge Engine
- Large-scale codebase comprehension using ultra-long context that retrieves 100,000 code files at once
- Building dedicated Agent teams with domain knowledge, task skills, and external tool interfaces for specific business scenarios
- Generating professional designs from voice input with QoderWork and delivering runnable, editable artifacts
- Terminal-native AI programming workflows with Qoder CLI
- Fully managed cloud AI agent deployment for multi-scenario integration
- Browser automation where the Jev judgment model drives next-step selection
Pros
- Team knowledge engine enables cross-session and cross-member knowledge reuse
- Ultra-long context retrieves 100,000 files at once, with industry-leading code comprehension
- Cross-project parallel tasks support multiple Agents working simultaneously, with a single screen for global control
- Quest mode independently developed, automating the entire process from goal definition to delivery
- Rich product line covers programming, design, terminal, and cloud scenarios, with a complete ecosystem
- Jev judgment model integration cuts permission judgment latency in Auto Mode from 3.13 to 0.53 seconds and drives next-step selection in browser automation
Summary
Qoder is an Agentic AI programming platform designed for professional developers and teams, ideal for users seeking efficient automated development, cross-project collaboration, and knowledge reuse. Its core advantage lies in the concept of "autopilot-style programming," which leverages a team knowledge engine, ultra-long context, cross-project parallel tasks, and customizable Agent teams to elevate AI from a supporting tool to an autonomous development partner, significantly improving development efficiency and delivery quality. With the 2026/9/30 integration of the Jev judgment model into its Agent Harness, generative models handle understanding and planning while Jev makes bounded semantic judgments, cutting permission judgment latency in Auto Mode from 3.13 to 0.53 seconds and driving next-step selection in browser automation.
Version History
- Qoder launches Fast mode. (2026-10-10): Alibaba Qoder launches Fast mode, debuting on desktop version 0.4.4 and above, available simultaneously for Personal and Enterprise editions as well as CN and International editions. First response time is reduced from 8 seconds to 3 seconds, Credits consumption drops from 1.1x to 0.8x, and output quality remains unchanged. The CN edition adds this tier, while the International edition's existing Performance mode is automatically upgraded to Fast.
- Jev judgment model integration (2026-09-30): Qoder integrated Jev, a System One judgment model from TypeSafe AI, into its Agent Harness: generative models handle understanding and planning while Jev makes bounded semantic judgments, cutting permission judgment latency in Auto Mode from 3.13 to 0.53 seconds and driving next-step selection in browser automation