Nejumi 4 - BFCL - Irrelevance Detection: leaderboard
Berkeley Function Calling Leaderboard as adapted and sampled by Nejumi Leaderboard 4 (Weights & Biases Japan): accuracy at declining to call any function when none is relevant. Accuracy (%), higher is better; the overall accuracy is Nejumi 4 - GLP - Function Calling.
Metric: Accuracy (%). Source: nejumi.ai. 145 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Llama 4 Scout | 100 | #648 |
| 2 | Gemini 2.5 Flash | 96.67 | #255 |
| 3 | Phi-4 | 96 | #707 |
| 4 | gemma-4-E4B-it | 95 | #577 |
| 5 | DeepSeek V3.2 (Thinking) | 95 | #210 (DeepSeek V3.2) |
| 6 | Gemma 4 E2B (IT) | 95 | #779 |
| 7 | GPT-4.1 | 93.33 | #258 |
| 8 | Grok 3 Mini | 93.33 | #289 |
| 9 | GPT-5 (2025-08-07) (High) | 93.33 | |
| 10 | GPT-5 Nano (2025-08-07) (High) | 93.33 | |
| 11 | GPT-5.4 (2026-03-05) (xHigh) | 93.33 | #63 (GPT-5.4 (2026-03-05)) |
| 12 | llm-jp-4-32B-a3B-Thinking | 93.33 | #477 |
| 13 | GPT-5.1 (2025-11-13) (High) | 93.33 | #107 (GPT-5.1 (2025-11-13)) |
| 14 | GPT-OSS-Swallow-120B-RL-v0.1 (Thinking) | 93.33 | #309 (GPT-OSS-Swallow-120B-RL-v0.1) |
| 15 | llm-jp-4.1-33B (High) | 93.33 | #256 (llm-jp-4.1-33B) |
Interactive version: theaggregate.ai/benchmark?slug=nejumi-4-bfcl-irrelevance-detection · How It Works · Data refreshed daily, snapshot 2026-11-05.