Skip to content

Benchmark

BFCL Non-Live AST

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
1.15

Score distribution

77.5%
Mean
84%
Median
17.4%
Std dev
59.8%
10th percentile
88.8%
90th percentile
0%
Min
0%90.7%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.15 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
51command-r7bfc· Berkeley Function-Calling Leaderboard data81%
52GPT-5 nanoPrompt· Berkeley Function-Calling Leaderboard data80.8%
53Gemma 3 12B ITPrompt· Berkeley Function-Calling Leaderboard data79.4%
54Granite 3.1 8B Instructfc· Berkeley Function-Calling Leaderboard data78.3%
55GPT 5.2Prompt· Berkeley Function-Calling Leaderboard data78.3%
56Nova Microfc· Berkeley Function-Calling Leaderboard data74.1%
57Mistral Smallfc· Berkeley Function-Calling Leaderboard data73.6%
58GPT-4.1 nanofc· Berkeley Function-Calling Leaderboard data73%
59GPT-4.1 nanoPrompt· Berkeley Function-Calling Leaderboard data72.4%
60Qwen3 0.6Bfc· Berkeley Function-Calling Leaderboard data71.8%
61Qwen3 0.6BPrompt· Berkeley Function-Calling Leaderboard data70%
62GPT-5 minifc· Berkeley Function-Calling Leaderboard data69.8%
63Phi 4Prompt· Berkeley Function-Calling Leaderboard data69.6%
64GPT-5 miniPrompt· Berkeley Function-Calling Leaderboard data68%
65GPT-5 nanofc· Berkeley Function-Calling Leaderboard data68%
66Mistral Medium 3fc· Berkeley Function-Calling Leaderboard data67.4%
67Gemma 3 4B ITPrompt· Berkeley Function-Calling Leaderboard data61.1%
68Claude Sonnet 4.5Prompt· Berkeley Function-Calling Leaderboard data59.8%
69Claude Haiku 4.5Prompt· Berkeley Function-Calling Leaderboard data55.4%
70OpenAI o3fc· Berkeley Function-Calling Leaderboard data40.4%
71Llama 3.2 1B Instructfc· Berkeley Function-Calling Leaderboard data38.4%
72o4 Minifc· Berkeley Function-Calling Leaderboard data37.7%
73Qwen3 235B A22B Instruct 2507fc· Berkeley Function-Calling Leaderboard data37.4%
74DeepSeek V3.2 Expfc· Berkeley Function-Calling Leaderboard data34.9%
75Gemma 3 1b itPrompt· Berkeley Function-Calling Leaderboard data20.2%
76Llama 3.1 Nemotron Ultra 253B v1fc· Berkeley Function-Calling Leaderboard data0%

Showing 51–76 of 76 results.

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