Skip to content

Benchmark

BFCL Overall

Agentic benchmarks measure how well models carry out tasks that take several steps. Scores are percentages.

Models measured
49
Scores recorded
76
Sources
1
IQ weight
1.22

Score distribution

40.7%
Mean
37.6%
Median
16.2%
Std dev
23.9%
10th percentile
63%
90th percentile
7.2%
Min
7.2%77.5%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.22 for chat models, measured over 49 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
51Gemma 3 12B IToverall, Prompt· Berkeley Function-Calling Leaderboard data30.4%
52GPT-4.1 minioverall, Prompt· Berkeley Function-Calling Leaderboard data29.7%
53Gemma 3 27B IToverall, Prompt· Berkeley Function-Calling Leaderboard data29.5%
54Phi 4overall, Prompt· Berkeley Function-Calling Leaderboard data28.8%
55Qwen3 1.7Boverall, fc· Berkeley Function-Calling Leaderboard data28.4%
56Llama 4 Scoutoverall, fc· Berkeley Function-Calling Leaderboard data28.1%
57Gemini 2.5 Flash-Liteoverall, Prompt· Berkeley Function-Calling Leaderboard data28%
58GPT-5 minioverall, Prompt· Berkeley Function-Calling Leaderboard data27.8%
59Open Mistral Nemooverall, fc· Berkeley Function-Calling Leaderboard data27.6%
60GPT-5 nanooverall, Prompt· Berkeley Function-Calling Leaderboard data27.6%
61Granite 3.1 8B Instructoverall, fc· Berkeley Function-Calling Leaderboard data27.1%
62Nova 2 Liteoverall, fc· Berkeley Function-Calling Leaderboard data27.1%
63Llama 3.1 8B Instructoverall, Prompt· Berkeley Function-Calling Leaderboard data25.8%
64Claude Haiku 4.5overall, Prompt· Berkeley Function-Calling Leaderboard data25.3%
65Nova Prooverall, fc· Berkeley Function-Calling Leaderboard data25%
66Claude Sonnet 4.5overall, Prompt· Berkeley Function-Calling Leaderboard data24.9%
67GPT-4.1 nanooverall, Prompt· Berkeley Function-Calling Leaderboard data24.9%
68Qwen3 0.6Boverall, fc· Berkeley Function-Calling Leaderboard data23.9%
69Qwen3 0.6Boverall, Prompt· Berkeley Function-Calling Leaderboard data22.4%
70Nova Microoverall, fc· Berkeley Function-Calling Leaderboard data22.3%
71Llama 3.2 3Boverall, fc· Berkeley Function-Calling Leaderboard data22%
72Gemma 3 4B IToverall, Prompt· Berkeley Function-Calling Leaderboard data19.6%
73Open Mistral Nemooverall, Prompt· Berkeley Function-Calling Leaderboard data19.3%
74Llama 3.2 1B Instructoverall, fc· Berkeley Function-Calling Leaderboard data10.8%
75Llama 3.1 Nemotron Ultra 253B v1overall, fc· Berkeley Function-Calling Leaderboard data10%
76Gemma 3 1b itoverall, Prompt· Berkeley Function-Calling Leaderboard data7.2%

Showing 51–76 of 76 results.

Data credits: Berkeley Function-Calling Leaderboard (Gorilla, UC Berkeley), Apache-2.0 (licence)