CursorProvisional

Composer 2.5

Composer / Composer 2.5 · composer-2-5

Cursor's own model for agentic coding workflows, tracked for Cursor-specific agent signals.

Objective score

26.9

Confidence

70%

Price evidence

Pricing evidence pending

Support this model Compare trend Official source

Other primary tiers in this family

Composer 1Composer 2.5

Scenario evidence

Six task-specific rankings

Feature development#43Obj. 26.90 · 0%Debug and repair#59Obj. 26.90 · 0%Refactoring#59Obj. 27.90 · 0%Code review#59Obj. 27.40 · 0%Large-repo agent work#33Obj. 29.40 · 0%Frontend / UI#59Obj. 27.90 · 0%

Evidence state

Signal coverage

Benchmarks: 0
Rank points: 0
Usage: 0
Agent: 1
Sources: 1

Objective evidence

Benchmarks and public signals

Swebench: —
Livecodebench: —
Aider: —
Artificial Analysis: —
Openrouter Usage: —
Kilo Usage: —

Persistent history

Recent ranking points

Evidence is being collected. Missing data is not treated as zero.

Usage signals

Evidence is being collected. Missing data is not treated as zero.

Agent and tool signals

Cursor CLI · 38.2

Cursor CLI - Composer 2.5 Fast

Source trail

Official Cursor model catalog Artificial Analysis Coding Agents · ok