Composer 2 AI Model Targets Long-Running Autonomous Coding Tasks
Cursor has launched Composer 2, a new proprietary AI coding model built to handle autonomous, extended programming workflows. Developed by San Francisco-based Anysphere, the model is positioned as a direct challenger to coding systems from OpenAI and Anthropic. The release comes at a high-stakes moment for the company, as it is reportedly in discussions to raise new funding at a valuation of about $50 billion.
Composer 2 Is the Third Generation of Cursor’s In-House Coding Model
Composer 2 marks the third version of Cursor’s internal model line since October 2025, following the original Composer and Composer 1.5. That release pace signals an aggressive push to improve model quality and strengthen Cursor’s position in the fast-moving AI coding market.
Cursor Says Composer 2 Delivers Frontier-Level Coding Performance
According to the company, Composer 2 reaches frontier-level results on coding benchmarks such as Terminal-Bench 2.0 and SWE-bench Multilingual. Cursor says the model was trained with reinforcement learning on long-horizon coding tasks that require hundreds of sequential actions, a setup aimed at improving performance on more complex and sustained software development work rather than isolated code completions.
Composer 2 Pricing Positions Cursor as a Faster and Cheaper Coding Agent
Cursor is also competing on cost and speed. Composer 2 is priced at $0.50 per million input tokens and $2.50 per million output tokens. A faster variant is available at $1.50 per million input tokens and $7.50 per million output tokens. That pricing structure suggests Cursor is trying to balance high-end coding performance with practical deployment options for teams that care about both latency and budget.
Continued Pretraining Helped Improve Composer 2 Model Quality
Cursor attributes the quality jump in Composer 2 to its first continued pretraining run. The company says this created a stronger base model that could scale more effectively through reinforcement learning. In simple terms, the model appears to have benefited not just from post-training optimization, but from deeper foundational improvements before reinforcement learning was applied.
Benchmark Scores Show Composer 2 Ahead of Anthropic Models
According to a LinkedIn post by analyst Michael Spencer, Composer 2 scored 60.0 on CursorBench and 59.8 on Terminal-Bench 2.0. Those figures reportedly place it ahead of Anthropic’s Opus 4.6 and Sonnet 4.5 on those benchmarks. That matters because benchmark leadership remains one of the clearest ways AI companies try to prove technical credibility in an increasingly crowded field.
Cursor Growth Shows Strong Demand for AI Coding Tools
Cursor’s launch of Composer 2 lands as the company continues to post rapid adoption and revenue growth. The startup says it now has more than one million daily active users and over seven million monthly active users, numbers that point to substantial usage across both individual developers and larger organizations.
Enterprise Adoption Is Becoming a Major Revenue Driver
The company says Stripe uses Cursor across more than 3,000 developers, and Cursor also claims usage by more than half of the Fortune 500. That kind of enterprise penetration matters because corporate adoption tends to be stickier, more lucrative, and more influential than individual experimentation.
Bloomberg reported that Cursor’s annualized revenue surpassed $2 billion in February, doubling in just three months. Roughly 60 percent of that revenue now comes from corporate customers, showing that enterprise demand is becoming central to the business rather than just an added layer on top of consumer growth.
Cursor Funding Talks Reflect Investor Confidence in AI Coding Startups
Cursor’s momentum is also showing up in its fundraising profile. The company was previously valued at $29.3 billion after raising $2.3 billion in a round co-led by Accel and Coatue in November 2025. It is now reportedly in early talks for another financing round at a valuation of around $50 billion.
A Higher Valuation Would Reinforce Cursor’s Position in the AI Developer Tools Market
If that new round happens near the reported figure, it would mark a significant valuation jump in a short time. That kind of move would underline how aggressively investors are backing AI infrastructure and developer tooling, especially companies that show both technical differentiation and strong enterprise revenue growth.
AI Coding Competition Is Intensifying Across the Market
Cursor is growing fast, but the pressure around it is obvious. The AI coding tools market has become one of the most contested areas in software, with major players spending heavily to win developers and enterprise buyers.
OpenAI, Anthropic, Google, and Microsoft Are All Expanding AI Coding Products
OpenAI acquired Windsurf for roughly $3 billion and continues to build its Codex product. Anthropic has been pushing Claude Code, which has reportedly drawn some individual developers away from Cursor. Google previewed Antigravity, a free AI IDE, earlier this year. Microsoft’s GitHub Copilot remains deeply entrenched in the market and recently expanded its enterprise offering by adding Anthropic and Google models.
Cursor’s Next Phase Depends on Whether Composer 2 Can Keep Up
Cursor’s ability to maintain its momentum will likely depend on whether its proprietary model strategy can keep pace with competitors backed by some of the biggest balance sheets in tech. Composer 2 is now the centerpiece of that strategy. If it can continue delivering strong benchmark performance, attractive pricing, and reliable long-range coding execution, it could help Cursor defend and expand its position in a market that is getting more crowded by the month.