K2-V2 (Medium) — benchmark results
Provider: LLM360. Released 2026-01-27. Access: Open.
Unified ELO 1530 ± 34, rank #723 of 1841 rated models, from 53 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AA IFBench | 55.1 | Accuracy (%) | 66.4 |
| AA Omniscience - Software Engineering (SWE) - Kotlin | 24 | Accuracy (%) | 63.4 |
| AA Omniscience - Software Engineering (SWE) - Java | 21 | Accuracy (%) | 63.1 |
| AA AIME 2025 | 64.67 | Accuracy (%) | 60.4 |
| AA LiveCodeBench | 54.07 | Pass@1 (%) | 60.2 |
| AA Omniscience - Software Engineering (SWE) - HTML | 34 | Accuracy (%) | 58.4 |
| AA MMLU-Pro | 76.07 | Accuracy (%) | 52.3 |
| AA Omniscience - Software Engineering (SWE) - TypeScript | 22.22 | Accuracy (%) | 49.6 |
| AA Omniscience - Software Engineering (SWE) - Swift | 36 | Accuracy (%) | 47.1 |
| AA-LCR | 28 | Score (self-reported) | 47.1 |
| AA Omniscience - Health | 18.3 | Accuracy (%) | 46 |
| AA Omniscience - Software Engineering (SWE) - R | 10 | Accuracy (%) | 44.9 |
Interactive version: theaggregate.ai/model?slug=k2-v2-medium · How It Works · Data refreshed daily, snapshot 2026-07-25.