On the stage of AI programming, Cursor's story is like a dramatic startup drama. Recently, the company released its latest self-developed model - Composer 2.5, making a strong comeback in the competitive market. This move marks Cursor's departure from its reliance on Claude and the start of its "self-developed counterattack," attracting widespread attention with a price far lower than competitors and performance comparable to theirs.

Over the past year, Cursor has faced many challenges, losing market share due to the rise of its competitor Claude Code. Claude Code rapidly increased its annualized revenue to $2.5 billion within just a few months, quickly capturing half of the industry. Cursor had to fight against its competitor, losing clients while still paying API call fees, which was frustrating. However, after a difficult period, Cursor finally delivered an impressive result with Composer 2.5.

Composer 2.5 is nearly equal in performance to Claude Opus 4.7, scoring 69.3% on terminal complex operations tests and 79.8% on cross-language engineering capability tests. More importantly, Cursor has significantly reduced the cost of task execution to less than $1 per task, far lower than the prices of Claude and GPT, which is essentially a "downward strike."

In terms of technology, Cursor has made deep optimizations in Composer 2.5, innovatively adopting a "working while being criticized" training approach to improve the model's real-time feedback capability. At the same time, through self-generated questions, the model demonstrated strong reverse analysis capabilities when facing challenges. Additionally, optimization of the underlying architecture has greatly improved training speed, reaching industry-leading levels.

Notably, Cursor's success also owes much to Elon Musk's attention. Before the release of Composer 2.5, Musk had announced a $6 billion acquisition option agreement with Cursor, although the terms were quite harsh. However, after the new model was released, Musk reversed his stance and personally supported Cursor, further enhancing its market position.

Cursor's strong comeback not only brought new life to itself but also injected new vitality into the AI programming industry. The future competition will be even more exciting, and Cursor's story continues.

Key Points:

🌟 **New Self-Developed Model**: Cursor launched Composer 2.5, freeing itself from reliance on Claude and returning to the competitive stage.

💰 **Ultra-Low Price Advantage**: Composer 2.5 costs less than $1 per task, far lower than Claude and GPT, offering high value for money.

🚀 **Enhanced Technical Strength**: Through real-time feedback and self-generated question training, Cursor has significantly improved model performance, achieving industry-leading training speeds.