BFCL V4 - Memory: leaderboard
Metric: Accuracy (%). Source: gorilla.cs.berkeley.edu. 109 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.5 (20251101) | 73.76 |
| 2 | Claude Sonnet 4.5 | 64.95 |
| 3 | Gemini 3 Pro (Preview) | 61.72 |
| 4 | Grok 4 | 55.91 |
| 5 | Claude Haiku 4.5 (20251001) | 54.41 |
| 6 | DeepSeek V3.2 Exp | 54.19 |
| 7 | Grok 4.1 Fast (Reasoning) | 53.98 |
| 8 | O3 (2025-04-16) | 51.83 |
| 9 | GPT-5.2 | 45.81 |
| 10 | Nanbeige3.5-Pro (Thinking) | 45.16 |
| 11 | GPT-5 Mini | 44.3 |
| 12 | Gemini 2.5 Flash | 41.29 |
| 13 | Nanbeige4-3B-Thinking-2511 | 36.77 |
| 14 | O4 Mini (2025-04-16) | 35.27 |
| 15 | Kimi K2 | 29.03 |
Interactive version: theaggregate.ai/benchmark?slug=bfcl-v4-memory · How It Works · Data refreshed daily, snapshot 2026-09-19.