BFCL V4 - Relevance Detection: leaderboard
Metric: Accuracy (%). Source: gorilla.cs.berkeley.edu. 109 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Llama 3.3 70B Instruct | 100 |
| 2 | Gemma 3 4B (IT) | 100 |
| 3 | GPT-4.1 | 100 |
| 4 | Llama 4 Scout Instruct | 100 |
| 5 | Llama 4 Maverick Instruct FP8 | 100 |
| 6 | Falcon3-7B-Instruct | 100 |
| 7 | Nanbeige3.5-Pro (Thinking) | 100 |
| 8 | Llama 3.1 8B Instruct | 93.75 |
| 9 | Gemma 3 12B (IT) | 93.75 |
| 10 | GPT-4.1 Nano | 93.75 |
| 11 | GPT-5 Nano | 93.75 |
| 12 | GPT-5 Mini | 93.75 |
| 13 | Qwen 3 8B | 93.75 |
| 14 | Qwen 3 30B A3B 2507 Instruct | 93.75 |
| 15 | O3 (2025-04-16) | 93.75 |
Interactive version: theaggregate.ai/benchmark?slug=bfcl-v4-relevance-detection · How It Works · Data refreshed daily, snapshot 2026-09-19.