Cursor launches fully self-trained model, Origin Git platform, and mobile app

Anysphere, a prominent startup in AI-assisted coding, revealed at its Compile26 company event that Cursor — its widely used IDE — will debut a fully self-trained AI foundation model. Alongside this, it introduced a new Git platform named Origin and a beta iOS version of its mobile app. These moves signal Cursor’s shift from a single code editor to a full developer ecosystem platform built for both human developers and AI agents.
Co-founder Michael Truell explained that the new model was trained entirely from scratch, marking a departure from the previous "Composer model" built on open-source foundations. Its parameter count rivals that of Anthropic's Claude Opus and OpenAI's GPT series, and its underlying compute capacity is 10 to 20 times greater than earlier Cursor models.
Notably, the model's use cases extend beyond coding, targeting broader non-programming tasks. On the infrastructure side, Anysphere was previously acquired by SpaceX, and the two companies jointly conducted the model's training. The training is still underway, with an official release expected in the coming weeks.
Alongside the core foundation model, Cursor also unveiled the next-generation Git platform Origin, designed specifically for collaboration between humans and AI agents. This platform is led by co-founder Tomas, who joined through the acquisition of Graphite, and runs on a brand-new cloud-native Git architecture. During extreme stress tests, Origin withstood thousands of AI agents simultaneously reading and writing to the same code repository, and demonstrated capabilities such as autonomously resolving merge conflicts, fixing CI test failures, and handling code comments. Origin is currently operational internally and with select partners, with a full launch planned for fall 2026.
To further expand the ecosystem, Cursor launched an iOS beta version of Cursor Mobile. This app enables users to remotely control and manage AI agents running locally or in the cloud from anywhere, view and comment on screenshots generated by the agent, and help resolve stuck tasks. From the fully self-trained foundation model to the Origin Git platform and now the mobile entry point, Cursor is reshaping the collaborative paradigm of software development. This shows that AI is not only generating code but also accelerating the redesign of underlying version control systems and development workflows, marking a critical phase in the engineering of large-scale AI agent collaboration.
Related article
Swedish AI Startup Lovable Eyes $13.2 Billion Valuation After Major Funding Round
As AI-driven coding tools gain traction, Swedish startup Lovable has secured a major funding round. The company aims to raise $3 billion, potentially boosting its valuation to $13.2 billion—double the $6.6 billion recorded last December. Menlo Ventur
Google Tests Remy AI Agent for Gemini as Focus Shifts to User Control
According to Business Insider, Google is testing Remy, a new AI personal agent for Gemini. This tool aims to execute tasks on behalf of users, streamlining both professional workflows and daily routines.Currently, Remy is undergoing testing in an int
How to fix Core Web Vitals for better SEO rankings
Streamline Report Card Comments with AI ToolsIntroductionAI Tools for Generating Report Card CommentsMagic SchoolAlmanac AIChat GPTUsing Magic School to Generate Report Card CommentsLogging into Magic SchoolSelecting the Report Card Comments ToolCust
Related Special Topic Recommendations
Comments (1)
0/500
Finally, a model that doesn't rely on someone else's infrastructure? 🤔 This feels like a massive shift for Cursor. I've been waiting for them to own the stack completely. The mobile app is a nice touch, but I'm more curious about how Origin handles collaboration. Will it actually solve the Git merge conflict nightmare, or is it just another shiny toy? 🛠️ Let's see if the self-trained model beats Claude or GPT-4 in real-world debugging scenarios. Fingers crossed it's not just marketing hype! 🚀

Anysphere, a prominent startup in AI-assisted coding, revealed at its Compile26 company event that Cursor — its widely used IDE — will debut a fully self-trained AI foundation model. Alongside this, it introduced a new Git platform named Origin and a beta iOS version of its mobile app. These moves signal Cursor’s shift from a single code editor to a full developer ecosystem platform built for both human developers and AI agents.
Co-founder Michael Truell explained that the new model was trained entirely from scratch, marking a departure from the previous "Composer model" built on open-source foundations. Its parameter count rivals that of Anthropic's Claude Opus and OpenAI's GPT series, and its underlying compute capacity is 10 to 20 times greater than earlier Cursor models.
Notably, the model's use cases extend beyond coding, targeting broader non-programming tasks. On the infrastructure side, Anysphere was previously acquired by SpaceX, and the two companies jointly conducted the model's training. The training is still underway, with an official release expected in the coming weeks.
Alongside the core foundation model, Cursor also unveiled the next-generation Git platform Origin, designed specifically for collaboration between humans and AI agents. This platform is led by co-founder Tomas, who joined through the acquisition of Graphite, and runs on a brand-new cloud-native Git architecture. During extreme stress tests, Origin withstood thousands of AI agents simultaneously reading and writing to the same code repository, and demonstrated capabilities such as autonomously resolving merge conflicts, fixing CI test failures, and handling code comments. Origin is currently operational internally and with select partners, with a full launch planned for fall 2026.
To further expand the ecosystem, Cursor launched an iOS beta version of Cursor Mobile. This app enables users to remotely control and manage AI agents running locally or in the cloud from anywhere, view and comment on screenshots generated by the agent, and help resolve stuck tasks. From the fully self-trained foundation model to the Origin Git platform and now the mobile entry point, Cursor is reshaping the collaborative paradigm of software development. This shows that AI is not only generating code but also accelerating the redesign of underlying version control systems and development workflows, marking a critical phase in the engineering of large-scale AI agent collaboration.
Swedish AI Startup Lovable Eyes $13.2 Billion Valuation After Major Funding Round
As AI-driven coding tools gain traction, Swedish startup Lovable has secured a major funding round. The company aims to raise $3 billion, potentially boosting its valuation to $13.2 billion—double the $6.6 billion recorded last December. Menlo Ventur
How to fix Core Web Vitals for better SEO rankings
Streamline Report Card Comments with AI ToolsIntroductionAI Tools for Generating Report Card CommentsMagic SchoolAlmanac AIChat GPTUsing Magic School to Generate Report Card CommentsLogging into Magic SchoolSelecting the Report Card Comments ToolCust
Finally, a model that doesn't rely on someone else's infrastructure? 🤔 This feels like a massive shift for Cursor. I've been waiting for them to own the stack completely. The mobile app is a nice touch, but I'm more curious about how Origin handles collaboration. Will it actually solve the Git merge conflict nightmare, or is it just another shiny toy? 🛠️ Let's see if the self-trained model beats Claude or GPT-4 in real-world debugging scenarios. Fingers crossed it's not just marketing hype! 🚀





Home






