Benchmark · agentic
CursorBench
Cursor's private, version-tagged internal agent eval built from real Cursor sessions; the trailing number is a version, not a score. Not one of the public suites it is quoted beside.
| Tanggal | Model | Skor | Sumber |
|---|---|---|---|
| 2026-09-01 | Claude Fable 5.1 | 73.4% | Anthropic merilis Claude Fable 5.1 dengan skor 52,6% di Terminal-Bench-Science dan bacaan cache yang lebih murah |
| 2026-08-13 | Grok 4.6 | 69.9% | SpaceXAI merilis Grok 4.6 dengan konteks 500K dan penalaran agentic yang lebih baik |
| 2026-07-31 | Claude Opus 5 | 70.0% | Anthropic meluncurkan Claude Opus 5; model OpenAI menembus Hugging Face selama uji keamanan |