We go into detail about the infrastructure behind large scale training including the kernels we developed and open-sourced for the project. We also discuss distributed training and environment scaling for RL.
@cursor_ai
-
Reinforcement Learning Phase Critical for Model Performance
By
–
The reinforcement learning phase is critical for final performance. We discuss the algorithms we apply for this stage. We find that simple approaches often work best, and improve performance broadly.
-
Continued Pretraining Boosts Downstream Coding Performance
By
–
We show how continued pretraining results in consistent improvements in downstream coding performance.
-
Composer 2 Uses Continued Pretraining and RL for Coding
By
–
Composer 2 had three main efforts: continued pretraining, reinforcement learning, and benchmark development. The goal of each was to closely emulate the Cursor environment to produce a highly intelligent coding model.
-
Composer 2 Training Technical Report Released
By
–
We're releasing a technical report describing how Composer 2 was trained.
-
Cursor Now Creates Figma Components Using Your Design System
By
–
Cursor can now create new components and frontends in Figma using your team's design system. pic.twitter.com/iHG2xk3Wj2
— Cursor (@cursor_ai) 24 mars 2026Cursor can now create new components and frontends in Figma using your team's design system.
-
Cursor Builds Instant Grep to Search Millions of Files in Milliseconds
By
–
Cursor can now search millions of files and find results in milliseconds. This dramatically speeds up how fast agents complete tasks. We're sharing how we built Instant Grep, including the algorithms and tradeoffs behind the design.
-
Cursor Launches Early Alpha of New Glass Interface
By
–
We're also sharing an early alpha of our new interface. https://
cursor.com/glass -
Continued Pretraining Boosts Model Quality and Cuts Serving Costs
By
–
We were able to significantly improve the model quality and cost to serve. These quality improvements come from our first continued pretraining run, providing a far stronger base to scale our reinforcement learning.
-
Frontier-Level Coding Model Pricing Tiers Revealed
By
–
It's frontier-level at coding, priced at: – Standard: $0.50/M input and $2.50/M output
– Fast: $1.50/M input and $7.50/M output