HOST DIAGNOSTIC · Other
Cursor Composer 1.5
Cursor Composer 1.5· Other· RESTRICTED
ACTIVE
HOST IDENTIFICATION
Other
Cursor Composer 1.5
ACTIVE
AVG SCORE
[11.2]
CAPABLE
STRONGEST
Speed
[16]
WEAKEST
Multimodal Fluency
[8]
MAXED (20/20)
0
of 20 attributes
20-DIMENSION EVALUATION
ATTRIBUTE SCORES · 20 DIMENSIONS
Cognitive
Bulk Apperception
10
Reasoning
11
Mathematical Precision
11
World Knowledge
10
Scientific Acumen
10
Technical
Code Generation
12
Tool Use
12
Multimodal Fluency
8
Speed
16
Cost Efficiency
12
Behavioral
Candor
12
Creativity
9
Tenacity
11
Self-Correction
11
Calibration
11
Operational
Instruction Following
12
Context Fidelity
11
Multi-turn Coherence
12
Planning and Execution
12
Safety Alignment
11
SCORING METHODOLOGY
CALIBRATED. Anysphere/Cursor's 2nd in-house coding model, released February 2026 (interim step between Composer 1 and Composer 2). Benchmarks (Cursor-published): CursorBench 44.2, Terminal-Bench 2.0 47.9%, SWE-Bench Multilingual 65.9% — a clear tier below Composer 2 (61.3/61.7/73.7). PRICING WAS THE WEAK POINT: $3.50/$17.50 per M tokens (Composer 2 later undercut it ~86%) — hence Cost Efficiency 13 (LOW; the only Composer scored low on cost, because at launch it genuinely was expensive). DIMENSION RATIONALE: Code Gen 13 (CursorBench 44.2 = clear step down from Composer 2's 16); Speed 17 (fast, the Composer signature); deliberately LOW generalist dims (coding specialist). Tool Use 13 (Cursor IDE only). Largely superseded by Composer 2 within ~5 weeks. Sources: VentureBeat (Composer 2 launch, citing 1.5 comparatives), buildfastwithai, RockB guide. Anchored against Cursor Composer 2 + 2.5.