2 min lesson
What “state-of-the-art coding agent” means at Cursor
Start at the first move in "What 'state-of-the-art coding agent' means at Cursor" and carry it through to the proof.
Step 1 of 3
“state-of-the-art coding agent” is not an abstraction here. It maps to specific shipped systems and the team's published research describes how they were built. Showing fluency with the actual products is a real differentiator, because most candidates discuss RL in general terms.
Two product surfaces anchor the work and they pose different RL problems.
Interactive diagram. Tab through its regions; each focused region shows its detail in the panel below.
Top to bottom: research here owns the systems, not just the algorithms.
Learn more
Full explanation
Composer 2: base-model selection
- Base chosen
- A strong open model (reported as Kimi K2.5The open base model Cursor continued-trained into Composer 2 (1T parameters, 32B active, 256K context), chosen mainly for how well it fit Cursor's serving infrastructure. Press Enter for the full definition.)
- Scale
- ~1T total parameters, ~32B active (a sparse MoE)
- Why it matters
- Training on top of a capable base saves enormous compute vs. pretraining from scratch
- The tradeoff
- You inherit the base's strengths and its quirks; the RL recipe must work with both
Treat these specifics as reported figures, not gospel - but know the shape of the decision.