NexusRaven-V2-13B: benchmark results
Nexusflow's NexusRaven-V2-13B, a function-calling LLM built on CodeLlama-13B-Instruct, supporting single, nested and parallel function calls with optional explanations, trained without any proprietary-LLM-generated data. Provider: Nexusflow. Released 2023-12-04. Access: Open.
Aggregate Rating 1352 ± 27, rank #1382 of 1656 rated models, from 29 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Nexus Function Calling - CVECPE Multi APIs | 25 | Function Call Accuracy (%) | 100 |
| Nexus Function Calling - Climate | 70.21 | Function Call Accuracy (%) | 100 |
| Nexus Function Calling - Heldout Combined | 59.9 | Function Call Accuracy (%) | 100 |
| Nexus Function Calling - Overall | 58.2 | Mean Nine-Task Function Call Accuracy (%) | 100 |
| Nexus Function Calling - Places API | 50 | Function Call Accuracy (%) | 100 |
| Nexus Function Calling - VT Multi Dependency | 38.78 | Function Call Accuracy (%) | 100 |
| Nexus Function Calling - VT Multi Disconnected | 42.86 | Function Call Accuracy (%) | 100 |
| Nexus Function Calling - OTX | 90.22 | Function Call Accuracy (%) | 83.3 |
| Enkrypt AI - Bias Risk | 77 | Risk Score | 69.3 |
| Nexus Function Calling - CVECPE | 66.67 | Function Call Accuracy (%) | 66.7 |
| Open LLM Leaderboard v1 - GSM8K | 20.92 | Accuracy (%) (5-shot) | 41.6 |
| Enkrypt AI - Jailbreak Risk | 10.8 | Risk Score | 40.8 |
Interactive version: theaggregate.ai/model?slug=nexusraven-v2-13b · How It Works · Data refreshed daily, snapshot 2026-10-20.