In January 2025, ByteDance released Trae, an AI-native IDE built on a VS Code kernel. The initial release supported only macOS, with a Windows version arriving a month later. It was a modest debut for a product that would, within eighteen months, become one of the most widely used AI coding tools in the world — and the centerpiece of a market that looks increasingly different from the one Cursor and Claude Code dominate in the West.
Western coverage of AI coding tools tends to begin and end with Cursor and Anthropic's Claude Code. That coverage is not wrong. Cursor has crossed $2 billion in annual recurring revenue. Claude Code's ARR rose from $17 million in April 2025 to $2.5 billion in February 2026. Both are formidable products with large, loyal user bases.
But they are not the whole market. In China, a parallel ecosystem has emerged — Trae from ByteDance, Qoder from Alibaba, and CodeBuddy from Tencent — each built around a different assumption about how developers actually work. The three tools are not clones of Cursor. They are responses to a different set of constraints: a developer population that writes in Chinese, a corporate market that demands private deployment and domestic model access, and an engineering culture that prioritizes full-pipeline delivery over individual coding efficiency.
Understanding that difference matters for anyone tracking the global AI coding market. The tools are not competing for the same users with the same features. They are competing on a different set of trade-offs.
Trae: The Volume Play
Trae is the most visible of the three, and the one with the clearest numbers.
By December 2025, Trae had reached 1.6 million monthly active users and more than 6 million total registered users, covering nearly 200 countries and regions. Its users generated nearly 100 billion lines of code in a single year, initiated close to 60 million sessions, and sent 500 million queries. Average weekly active days per user approached five — a retention metric that suggests habitual use rather than curiosity.
The numbers accelerated in 2026. At WAIC in July, ByteDance announced Trae IDE 3.0, with monthly active users surpassing 5 million, support for more than 50 programming languages, and a "Solo Mode" designed for unattended, end-to-end development — natural language in, deployed application out. Five enterprise customers reported saving 500 million yuan in development costs over a year, though the company has not disclosed the methodology behind that figure.
Trae's positioning shifted alongside its user base. The original product was an AI-native IDE with code completion, chat, and a Builder mode for generating full projects. The 3.0 release repositioned it as an end-to-end agent — a tool that could autonomously complete coding, testing, and deployment, running overnight and on weekends to process multiple projects.
That positioning is closer to Devin or Claude Code's terminal agent than to Cursor's editor-first experience. It is a bet that the next phase of AI coding is not about helping developers write code faster, but about removing the developer from the loop for well-defined tasks.
Trae's pricing reflects its volume strategy. The domestic version is free for individual users, with a token-based paid tier introduced in February 2026. The international version, available in more than 100 countries, operates on a separate account system and pricing structure. The free tier is a customer acquisition tool. The enterprise tier is where the revenue lives.
Qoder: The Market Share Leader
If Trae is the volume leader by users, Qoder is the volume leader by revenue.
According to IDC's 2025 China AI Coding Market Share Report, Alibaba's Qoder — formerly known as Tongyi Lingma — captured 47.6 percent of the market, more than the combined share of the second through fifth competitors. Zhipu's CodeGeeX held 11.5 percent, SenseTime's Raccoon 10.5 percent, Tencent's CodeBuddy 6.9 percent, and Baidu's Comate 6.0 percent.
The market itself is growing quickly. IDC estimated China's AI coding market at 399 million yuan in 2025, with a projected 1.173 billion yuan in 2026 — nearly triple the previous year. Qoder alone has more than 5 million users and serves hundreds of thousands of enterprises, including FAW Group, CITIC Securities, and AsiaInfo Technologies.
Qoder's product differs from Trae's in emphasis. Where Trae leans toward autonomous end-to-end delivery, Qoder is built around what Alibaba calls "agentic coding" — a layered capability that covers assisted coding, collaborative coding, and autonomous coding. Its Quest Mode allows an agent to clarify requirements, plan a solution, execute code, and verify results without continuous human intervention.
Qoder's distribution advantage is its integration with Alibaba Cloud. The tool is available as a desktop IDE, a CLI, and a JetBrains plugin, with pay-as-you-go, coding plan, and token plan options. For enterprises already on Alibaba Cloud, Qoder is the path of least resistance. For enterprises that require private deployment or domestic model access, it is often the only compliant option.
Qoder also has an international version, operated as a separate product with its own account system, credits, and model access. The international version added a "Sonus" model in September 2026, optimized for long-task execution and computer use, and a Fast Mode in October that improved generation speed while reducing token consumption.
CodeBuddy: The Full-Stack Pipeline
Tencent's CodeBuddy takes a third approach. It is less an IDE and more a pipeline — a tool designed to connect product requirements, design, development, and deployment into a single workflow.
CodeBuddy's feature set reflects that orientation. It can generate a structured product requirements document from a natural language description, convert hand-drawn sketches into high-fidelity design mockups, translate Figma designs into front-end and back-end code, and deploy the result to a sandbox environment — all within the same interface. Its Agent Teams feature allows multiple CodeBuddy instances to collaborate, with a leader agent assigning tasks and member agents executing them.
The product's deepest integration is with WeChat. CodeBuddy's intelligent code review agent supports batch review of WeChat Mini Program code, and the IDE includes one-click deployment to Tencent Cloud services. For developers building on Tencent's ecosystem, that integration is not a feature. It is the reason to choose the tool.
CodeBuddy is the smallest of the three by market share — 6.9 percent in IDC's 2025 ranking. But it is the most vertically integrated. Tencent's cloud, mini-program platform, and enterprise software stack give CodeBuddy a distribution channel that neither Trae nor Qoder can match for WeChat-native development.
The tool is available in both domestic and international versions. The international version supports overseas model access and offers a Pro tier, with pricing available on Tencent Cloud International's product page.
What the Comparison Reveals
The three Chinese tools are not competing with Cursor on Cursor's terms. They are competing on different axes.
Cursor's advantage is the individual developer experience. Its Composer mode, multi-file diff, and codebase indexing are widely regarded as the benchmark for AI-native editing. Its pricing — $20 per month for Pro, with annual billing at $16 per month — is designed for individual adoption. It is a tool that makes a single developer faster.
The Chinese tools' advantage is the organizational pipeline. Trae's Solo Mode removes the developer from well-defined tasks. Qoder's Quest Mode and enterprise integration are built for large codebases and Alibaba Cloud environments. CodeBuddy's product-to-deployment workflow is designed for teams that need a single tool to cover the entire development lifecycle.
That difference reflects the markets each tool serves. Cursor's core users are individual developers and small teams in markets where the primary constraint is coding speed. Trae, Qoder, and CodeBuddy's core users are enterprises in markets where the primary constraints are data compliance, domestic model access, and integration with existing cloud and enterprise software stacks.
The Chinese tools are also priced differently. Trae's domestic version is free for individuals. Qoder offers pay-as-you-go and token-based plans. CodeBuddy offers tiered plans through Tencent Cloud International. Cursor's Pro tier is $20 per month. The price differences reflect different assumptions about willingness to pay, but they also reflect different cost structures — the Chinese tools are optimized for domestic inference costs and token pricing, not for the premium margins of a closed-model API.
The Market That Western Coverage Misses
The Chinese AI coding market is smaller than the global market. IDC's estimate of 1.173 billion yuan for 2026 is roughly $165 million at current exchange rates — a fraction of the $7.37 billion global market.
But the growth rate is higher. The Chinese market is projected to grow at a 40 percent compound annual rate through 2030, compared with 26.6 percent globally. And the competitive dynamics are different. The top player holds nearly half the market. The second through fifth players are fighting for single-digit shares. The market is consolidating around a few large platforms, each backed by a cloud provider with its own enterprise customer base.
That structure produces different product priorities. In a market where the top player has 47.6 percent share and the distribution channel is the cloud platform, the winning strategy is not to build the best individual developer tool. It is to build the tool that integrates most deeply with the enterprise's existing cloud, data, and deployment infrastructure.
That is what Qoder, Trae, and CodeBuddy are doing. They are building the AI coding layer for their respective cloud ecosystems — and in doing so, they are creating a market that operates on different rules.
What to Watch
The Chinese AI coding tools are at an earlier stage of international expansion than Cursor or Claude Code. Trae has the largest overseas presence, with users in nearly 200 countries and regions. Qoder and CodeBuddy have international versions but smaller global footprints.
The question is whether the model travels. A tool optimized for WeChat Mini Program development or Alibaba Cloud integration has limited appeal outside those ecosystems. A tool optimized for autonomous end-to-end delivery — Trae's Solo Mode — is more portable, but it competes directly with Claude Code and Devin in a market where those products already have momentum.
What the Chinese tools demonstrate is that there is more than one way to build an AI coding product. Cursor's editor-first approach is one path. Trae's agent-first approach is another. Qoder's enterprise-integration approach is a third. CodeBuddy's pipeline approach is a fourth.
Western coverage has largely focused on the first. The other three are building a market that is growing faster, consolidating faster, and integrating more deeply with the infrastructure that enterprises already use. That market is worth watching — not because it will replace Cursor, but because it shows what AI coding looks like when the starting constraint is not individual productivity, but organizational delivery.
Sources: Trae official 2025 Product Report via TRAE Forum (February 28, 2026); Huawei Cloud Blog (July 28, 2026); 36Kr EU "Big Techs' Covert Battle for the 100-Billion-Yuan Market" (July 30, 2026); IDC 2025 China AI Coding Market Share Report via Shanghai Securities News (July 16, 2026); 36Kr "Cursor: Once Half of Anthropic's Revenue" (June 16, 2026); PingWest "How to Choose an AI Coding Platform" (October 15, 2026); Tencent Cloud International — CodeBuddy Product Page (August 7, 2026); Cursor Pricing Page (2026).
Disclaimer
The information provided in this article is for general informational and educational purposes only. It does not constitute legal, financial, or professional advice. The author and publisher are not responsible for any actions taken based on the content of this article. Readers should consult qualified professionals for advice specific to their situation. All trademarks and references to third-party products, services, or organizations are the property of their respective owners. The performance data and benchmarks discussed are based on specific research studies and may not generalize to all use cases or environments. As of the publication date, the AI landscape continues to evolve rapidly, and readers should verify current information independently.
Limitations
This analysis is based on vendor disclosures, analyst reports, and public reporting available as of the article date.
Market share and revenue figures come from third-party estimates (IDC, company announcements); independent verification was not performed.
User counts are company-reported and may count registered accounts rather than active paying users.
Pricing and product tiers change frequently and differ by region.
Comparative claims rely on public reporting, not hands-on testing of every product.
Reporting standards differ across jurisdictions; figures may not be directly comparable.
This sector is changing quickly; claims can become outdated soon.
Sources
- Trae official 2025 Product Report via TRAE Forum (February 28, 2026)
- Huawei Cloud Blog (July 28, 2026)
- 36Kr EU "Big Techs' Covert Battle for the 100-Billion-Yuan Market" (July 30, 2026)
- IDC 2025 China AI Coding Market Share Report via Shanghai Securities News (July 16, 2026)
- 36Kr "Cursor: Once Half of Anthropic's Revenue" (June 16, 2026)
- PingWest "How to Choose an AI Coding Platform" (October 15, 2026)
- Tencent Cloud International — CodeBuddy Product Page (August 7, 2026)
- Cursor Pricing Page (2026).
The information provided in this article is for general informational and educational purposes only. It does not constitute legal, financial, or professional advice. The author and publisher are not responsible for any actions taken based on the content of this article. Readers should consult qualified professionals for advice specific to their situation. All trademarks and references to third-party products, services, or organizations are the property of their respective owners. The performance data and benchmarks discussed are based on specific research studies and may not generalize to all use cases or environments. As of the publication date, the AI landscape continues to evolve rapidly, and readers should verify current information independently.
Limitations: This analysis is based on vendor disclosures, analyst reports, and public reporting available as of the article date.; Market share and revenue figures come from third-party estimates (IDC, company announcements); independent verification was not performed.; User counts are company-reported and may count registered accounts rather than active paying users.; Pricing and product tiers change frequently and differ by region.; Comparative claims rely on public reporting, not hands-on testing of every product.; Reporting standards differ across jurisdictions; figures may not be directly comparable.; This sector is changing quickly; claims can become outdated soon.