Nejumi 4 - BFCL - Non-Live AST: leaderboard
Berkeley Function Calling Leaderboard as adapted and sampled by Nejumi Leaderboard 4 (Weights & Biases Japan): abstract-syntax-tree accuracy on the non-live simple and multiple-function categories. Accuracy (%), higher is better; the overall accuracy is Nejumi 4 - GLP - Function Calling.
Metric: Accuracy (%). Source: nejumi.ai. Saturation forecast: Not forecast. 145 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Grok 3 | 85.93 | #313 |
| 2 | Claude Sonnet 4 (Non-reasoning) | 84.07 | #207 (Claude Sonnet 4) |
| 3 | Gemini 2.5 Flash | 82.96 | #255 |
| 4 | Qwen 3 30B A3B 2507 (Thinking) | 82.96 | #379 |
| 5 | Claude Opus 4 (Non-reasoning) | 82.59 | #170 (Claude Opus 4) |
| 6 | Claude Opus 4.7 (xHigh) | 82.22 | #51 (Claude Opus 4.7) |
| 7 | GPT-4o (2024-11-20) | 81.85 | #392 |
| 8 | Claude Sonnet 4 (Thinking) | 81.85 | #207 (Claude Sonnet 4) |
| 9 | qwen2.5-bakeneko-32B-instruct-v2 | 81.85 | #570 |
| 10 | Gemini 3 Pro (Preview) | 81.48 | #71 |
| 11 | Qwen 3 235B A22B 2507 (Thinking) | 81.48 | #263 |
| 12 | GPT-OSS-Swallow-20B-RL-v0.1 (Thinking) | 81.48 | #451 (GPT-OSS-Swallow-20B-RL-v0.1) |
| 13 | O1 (2024-12-17) (High) | 81.48 | #157 (O1 (2024-12-17)) |
| 14 | Qwen 3 Max (Preview) | 81.11 | #248 |
| 15 | Qwen 3.5 122B A10B (Thinking) | 81.11 | #184 (Qwen 3.5 122B A10B) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=nejumi-4-bfcl-non-live-ast · How It Works · Data refreshed daily, snapshot 2026-11-05.