BFCL V4 - Web Search: leaderboard
Metric: Accuracy (%). Source: gorilla.cs.berkeley.edu. 109 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.5 (20251101) | 84.5 |
| 2 | Claude Haiku 4.5 (20251001) | 83.5 |
| 3 | Grok 4.1 Fast (Reasoning) | 82.5 |
| 4 | Grok 4 | 82 |
| 5 | GPT-5 Mini | 82 |
| 6 | Claude Sonnet 4.5 | 81 |
| 7 | Gemini 3 Pro (Preview) | 80 |
| 8 | O3 (2025-04-16) | 77 |
| 9 | GPT-5.2 | 75.5 |
| 10 | O4 Mini (2025-04-16) | 75.5 |
| 11 | Grok 4.1 Fast (Non-reasoning) | 75 |
| 12 | GPT-5 Nano | 72.5 |
| 13 | DeepSeek V3.2 Exp | 69.5 |
| 14 | GPT-4.1 | 68 |
| 15 | Kimi K2 | 66.5 |
Interactive version: theaggregate.ai/benchmark?slug=bfcl-v4-web-search · How It Works · Data refreshed daily, snapshot 2026-09-19.