Qoder
Alibaba's agentic IDE, which absorbed Tongyi Lingma (通义灵码) as Qoder CN on 20 May 2026. Repo Wiki and Quest Mode on credit-based pricing, with Linux builds since version 1.0.
Overview
The naming needs sorting out before anything else. Alibaba launched Qoder in August 2025 as an agentic IDE, reached 100,000 users inside five days and passed 5 million globally by mid-2026. Then on 20 May 2026 it retired the Tongyi Lingma brand and relaunched that product as Qoder CN. Same family name, different product. Qoder CN runs GLM, DeepSeek, Kimi and MiniMax on infrastructure inside China, aimed at banks and government buyers who need data to stay onshore. The international Qoder routes across Qwen, Claude, Gemini and GPT. Accounts and model catalogues do not overlap.
Version 1.0 arrived on 15 May 2026 and added Linux alongside the existing Windows and macOS clients. Two features carry the product. Repo Wiki indexes a repository into generated documentation and Knowledge Cards so the agent stops re-reading the same files on every task, although that index lives inside Qoder rather than in your working tree, which some security teams will object to. Quest Mode takes a written specification and runs it asynchronously across multiple files, coming back only when it hits a decision it cannot make. Both depend heavily on how well you write the spec, which is a skill many teams have not built yet.
Pricing is credit-based and that is where costs escape. Free gives a two-week Pro trial with 300 credits, unlimited completions and next edits on every tier, plus bring-your-own-key. Pro is $20 a month for 2,000 premium credits, Pro+ $60 for 6,000 and Ultra $200 for 20,000. Autonomous Quest runs consume that quota quickly and certain built-in agent calls still bill even against your own key. Indexing performance also falls off on large monorepos. For an engineer outside China the international build is straightforward: English end to end, standard desktop installers on all three platforms and no mainland cloud account in the path.
Key Features
- ✓ Qoder 1.0 shipped 15 May 2026, adding Linux to the existing Windows and macOS builds
- ✓ Repo Wiki generates living project documentation and Knowledge Cards from the codebase, held in Qoder's own index rather than written into your working tree
- ✓ Quest Mode takes a written specification and executes it asynchronously, returning only when blocked or when a decision is needed
- ✓ Model routing across Qwen, Claude, Gemini and GPT on the international build, with unlimited code completion on every plan including Free
- ✓ Bring-your-own-key supported down to the Free tier, alongside a 14-day Pro trial with 300 credits
- ✓ Qoder CN, the renamed Tongyi Lingma, runs GLM, DeepSeek, Kimi and MiniMax on China-hosted infrastructure for banking and government buyers with data residency requirements
Where it holds
- • English interface and English documentation throughout the international build, which is not something every mainland dev tool manages
- • Repo Wiki is a genuinely distinct idea: persistent generated project knowledge the agent reads instead of re-scanning files every session
- • Free tier is unusually generous, with unlimited completions and next edits rather than a hard monthly cap
- • Reached 5 million users globally by mid-2026 from an August 2025 launch, so the product has the traffic to keep shipping
Where it breaks
- • Repository indexing degrades on large codebases, with reported multi-hour partial indexing above roughly 6,000 files
- • Credit consumption on autonomous Quest runs is difficult to predict, and some built-in agent capabilities still bill credits even when you supply your own API key
- • Proprietary with no self-hosted option and no documented path for running a local model as the primary engine
- • Two separate products carry the name. Qoder CN is the renamed Tongyi Lingma with a Chinese model catalogue on mainland infrastructure, and accounts do not carry across from the international build
My Take
On 20 May 2026 the Tongyi Lingma brand disappeared into Qoder CN, and that single rename explains most of the confusion in English write-ups still comparing the two as separate tools. Version 1.0 had landed five days earlier with Linux builds alongside Windows and macOS. Repo Wiki is the feature that earns the subscription, generating persistent project documentation the agent reads rather than re-scanning your tree each session, though indexing crawls once a repository passes roughly 6,000 files. At $20 a month for 2,000 credits it is priced directly against Cursor, and Quest Mode will drain that quota faster than you expect.
Quick Info
- Pricing:
- freemium
- Openness:
- Proprietary
- Starting at:
- Free plan at $0 with a two-week Pro trial carrying 300 credits, unlimited code completions and next edits on every tier including Free, plus bring-your-own-key. Pro is $20 per month for 2,000 premium-model credits, Pro+ $60 for 6,000, Ultra $200 for 20,000. Teams is $40 per seat per month for 3,000 credits per seat. Credit packs add 1,500 credits for $20 with one-month validity. Free plans fall back to basic models once premium credits are exhausted rather than blocking.
- Added:
- Aug 2026
- Updated:
- Aug 2026
Use Cases
Judge it on your own work
The notes above say where Qoder holds and where it breaks. The fastest check is your own workload.
Visit website ↗Qoder compared head to head
Alternatives to Qoder
Cursor
freemiumA VS Code fork built around agentic editing, with its own Composer models, a pull request review bot and SOC 2 controls for teams that want AI in the editor.
Trae
freemiumByteDance's AI IDE and standalone SOLO agent that scaffolds projects, runs terminal commands and tests in a browser, at $10 a month for Pro against Cursor's $20.
Tencent CodeBuddy
freemiumTencent Cloud's coding assistant (腾讯云代码助手), a distinct product from the Hunyuan chatbot. Standalone IDE, editor plugins and a CLI with native Agent Client Protocol support.
GitHub Copilot
freemiumMicrosoft's AI coding assistant across VS Code, JetBrains and the GitHub web, with background agents that open pull requests and org-level policy controls.