Category ranking · 9 models · checked 2026-10-13
Small fast 'System 1'
Ranked by BFCL V4, overall accuracy. Source: Berkeley Function-Calling Leaderboard data_overall.csv.
BFCL V4, overall accuracy
3 with a published score, 6 not yet measured.
# | Model | Parameters | Score | Source | Licence |
|---|---|---|---|---|---|
| 1 | Qwen3 0.6B Alibaba Qwen | 752 M | 23.93%Sourced | Source Qwen3-0.6B (FC) row, rank 92. | Apache-2.0 |
| 2 | Llama 3.2 1B Meta | 1.2 B | 10.82%Sourced | Source Llama-3.2-1B-Instruct (FC) row, rank 107. | Llama 3.2 Community |
| 3 | Gemma 3 1B Google | 1.0 B | 7.17%Sourced | Source Gemma-3-1b-it (Prompt) row, rank 109. | Gemma Terms of Use |
| — | Gemma 3 270M Google | 268 M | Not yet measured | — | Gemma Terms of Use |
| — | Granite 4.0 H Micro IBM | 3.2 B | Not yet measured | — | Apache-2.0 |
| — | LFM2 1.2B Liquid AI | 1.2 B | Not yet measured | — | LFM Open License v1.0 |
| — | Phi-4 mini Microsoft | 3.8 B | Not yet measured | — | MIT |
| — | Qwen3 1.7B Alibaba Qwen | 2.0 B | Not yet measured | — | Apache-2.0 |
| — | SmolLM3 3B Hugging Face | 3.1 B | Not yet measured | — | Apache-2.0 |
How to read this: one public benchmark per category, so you can compare like with like. We copy the published number and link it; we do not re-run it, average different benchmarks or fill gaps. Vendor-reported numbers use each vendor's own settings, so treat small gaps as ties. “Not yet measured” means we found no comparable public score, not that the model is weak.
Every model was checked against the Hugging Face API on 2026-10-13: it exists, is public, and has the parameter count and licence shown. Other categories: Math and reasoning · Coding · Vision-language · OCR and documents · Speech to text · Text to speech · Image generation · Embeddings and rerankers · Biomedical · Tool use and agents · Multilingual and translation · Time series · Robotics (VLA).
Machine-readable: /api/v1/categories/small-fast.json