A model customized for end-to-end programming in Qoder, built on Qwen-Coder with large-scale reinforcement learning for the Qoder Agent framework, tools, and scenarios. Outperforms Cursor Composer-1 in real-world task resolution, with 50% higher accuracy in Windows terminal commands.
Upgraded completion and NES to NEXT.
Transitions from passive completion to proactive prediction. Supports automatic dependency imports.
Supports manual triggering of NEXT code completion.
Full database support (@database).
Reference databases as context in Qoder Ask/Agent mode.
Generate and execute SQL in the Query Console.
AGENTS.md support: auto-detected and loaded from the project root.
Major model upgrade: Qwen-Coder-QoderA model customized for end-to-end programming, built on Qwen-Coder with large-scale reinforcement learning for the Agent framework, tools, and scenarios. Achieves excellent task resolution rates in real-world evaluations, with 50% higher accuracy in Windows terminal commands.
Quest autonomous programming (public beta)
Autonomous programming agent: Independently delivers end-to-end, high-quality results without continuous human intervention.
Requirement alignment: Clarifies tasks before execution through intent recognition, requirement clarification, and collaborative spec creation.
Long-running tasks: Runs continuously and supervises tasks to completion.
Autonomous quality assurance: Independently verifies and fixes deliverable quality.
Continual evolution: Actively evolves, remembers your style, and learns new technologies.
More scenarios: Quest recognizes intent and routes to the best-fit capability.
Spec-driven development: Align on requirements and constraints, then execute and accept clear, traceable results.
From idea to product: Build websites and prototypes from scratch with top-tier visual models for high design quality and usability.
Built-in planning agent: Collaborate on execution plans before complex tasks to reduce rework and improve code quality and controllability.
Custom extensions: Build intelligent workflows with custom agents (subagents), Skills, and Commands.
Multi-agent parallel sessions: Run multiple sessions in parallel to handle different tasks simultaneously.
Repo Wiki (public beta, Enterprise Dedicated Edition)
Auto-generates structured project documentation and continuously tracks code and document changes.
When querying knowledge, requesting code explanations, or adding features, combines project analysis with indexed context for accurate answers and documentation support, giving the agent deeper codebase understanding.
