BFCL (v4) leaderboard
As of November 14, 2026, among the results listed here, inclusionAI Ling 3.0 Flash has the highest reported BFCL (v4) score, 73, published by inclusionAI, ahead of Qwen3.7 Plus at 72.9. 6 models have a BFCL (v4) result in our records.
Scores as the model makers published them, not re-run by us; each row links to its announcement.
6 models ranked · 4 makers
| Rank | Model | Score | Setting | Reported by | Date |
|---|---|---|---|---|---|
| 1 | Ling 3.0 Flash | 73 | Not published | inclusionAI | July 23, 2026 (model release) · read from the maker's chart |
| 2 | Qwen3.7 Plus | 72.9 | Not published | Qwen | May 21, 2026 |
| 3 | LFM3-1.7B | 71.4 | Not published | Liquid | November 6, 2026 (model release) |
| 4 | Qwen3.6 Plus | 68.9 | Not published | Qwen | May 21, 2026 |
| 5 | LFM2.5-2.6BRetired | 55.9 | Not published | Liquid | November 6, 2026 (model release) |
| 6 | Granite 4.2 8B | 52.4 | Not published | IBM Granite | August 31, 2026 (model release) |
How much to trust these numbers
- Source
- Reported by the model makers themselves. Not re-run by LLMBenchmark or anyone independent.
- Settings
- All ranked results use the same reported setting.
- Spread
- 73 to 52.4 across 6 models.
- Freshness
- Newest result published November 6, 2026; oldest May 21, 2026.
- Charts
- Some values were read from charts in the announcements. Two independent readings had to match exactly.
What BFCL (v4) measures
BFCL v4 asks models to call tools accurately; higher scores indicate better tool-calling accuracy. Source: Berkeley Function Calling Leaderboard (BFCL) V4 - Gorilla