agents benchmark
CursorBench 4.0 Leaderboard
CursorBench 4.0 evaluates coding agents on long-horizon coding tasks from real Cursor sessions; version 4.0 adds longer-horizon tasks, and its scores are not comparable with CursorBench 3.2.
Updated Sep 24, 2026
Models2
Model coverage2
MetricScore
EvidenceB
CursorBench 4.0 Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelAN | Score57.80% | Percentile100.00% | Participants2 | EvidenceC | Evaluated |
| Rank02 | ModelXA | Score46.30% | Percentile0.00% | Participants2 | EvidenceC | Evaluated |
CursorBench 4.0 Highlights
The leading models and scores on this benchmark.
CursorBench 4.0 Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for CursorBench 4.0
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
What is CursorBench 4.0?
What CursorBench 4.0 measures and how its scores work.
CursorBench 4.0 is an agents benchmark based on long-horizon coding tasks from real Cursor sessions.
It reports a Score as a ratio for coding agents on these tasks.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- CursorBench 4.0
- Modality
- text
- Primary category
- agents
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- overall
LLM Stats. Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about CursorBench 4.0.
Which model scores highest on CursorBench 4.0?
Claude Opus 5.5 is currently ranked first with 57.80%.
What are the top three models on CursorBench 4.0?
The current leaders are Claude Opus 5.5 (57.80%) and Grok 4.7 (46.30%).
Which CursorBench 4.0 model has the lowest official input price?
Grok 4.7 has the lowest matched official input price at $2.0 input / $6.0 output per 1M tokens.
Which models are fastest among CursorBench 4.0 results?
The fastest matched records are Claude Opus 5.5 (64.14 tok/s via Anthropic) and Grok 4.7 (5.32 tok/s via xAI).
Does the highest score result prove overall model quality?
No. This benchmark measures one defined capability or task. The overall LLMBoard score uses a separate aggregation across eligible benchmark evidence.