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
1Claude Opus 4.5overall, fc· Berkeley Function-Calling Leaderboard data77.5%
2Claude Sonnet 4.5overall, fc· Berkeley Function-Calling Leaderboard data73.2%
3Gemini 3 Pro Previewoverall, Prompt· Berkeley Function-Calling Leaderboard data72.5%
4GLM 4.6overall, FC thinking· Berkeley Function-Calling Leaderboard data72.4%
5Grok 4.1 Fastoverall, fc· Berkeley Function-Calling Leaderboard data69.6%
6Claude Haiku 4.5overall, fc· Berkeley Function-Calling Leaderboard data68.7%
7Gemini 3 Pro Previewoverall, fc· Berkeley Function-Calling Leaderboard data68.1%
8OpenAI o3overall, Prompt· Berkeley Function-Calling Leaderboard data63.1%
9Grok 4overall, Prompt· Berkeley Function-Calling Leaderboard data63%
10Grok 4overall, fc· Berkeley Function-Calling Leaderboard data61.4%
11Kimi K2 Instructoverall, fc· Berkeley Function-Calling Leaderboard data59.1%
12Command Aoverall, fc· Berkeley Function-Calling Leaderboard data57.1%
13DeepSeek V3.2 Expoverall, Prompt + Thinking· Berkeley Function-Calling Leaderboard data56.7%
14Gemini 2.5 Flashoverall, fc· Berkeley Function-Calling Leaderboard data56.2%
15GPT 5.2overall, fc· Berkeley Function-Calling Leaderboard data55.9%
16GPT-5 minioverall, fc· Berkeley Function-Calling Leaderboard data55.5%
17DeepSeek V3.2 Expoverall, fc· Berkeley Function-Calling Leaderboard data54.1%
18GPT 4.1overall, fc· Berkeley Function-Calling Leaderboard data54%
19o4 Minioverall, fc· Berkeley Function-Calling Leaderboard data53.2%
20Qwen3 235B A22B Instruct 2507overall, Prompt· Berkeley Function-Calling Leaderboard data52.2%
21GPT-5 nanooverall, fc· Berkeley Function-Calling Leaderboard data51.5%
22Gemini 2.5 Flashoverall, Prompt· Berkeley Function-Calling Leaderboard data50.9%
23GPT-4.1 minioverall, fc· Berkeley Function-Calling Leaderboard data50.5%
24o4 Minioverall, Prompt· Berkeley Function-Calling Leaderboard data50.3%
25Qwen3 32Boverall, fc· Berkeley Function-Calling Leaderboard data48.7%
26OpenAI o3overall, fc· Berkeley Function-Calling Leaderboard data48.6%
27Qwen3 235B A22B Instruct 2507overall, fc· Berkeley Function-Calling Leaderboard data48%
28Qwen3 32Boverall, Prompt· Berkeley Function-Calling Leaderboard data46.8%
29GPT 5.2overall, Prompt· Berkeley Function-Calling Leaderboard data45.3%
30Qwen3 8Boverall, fc· Berkeley Function-Calling Leaderboard data42.6%
31Qwen3 30B A3B Instructoverall, fc· Berkeley Function-Calling Leaderboard data41.4%
32Qwen3 14Boverall, fc· Berkeley Function-Calling Leaderboard data41%
33Qwen3 8Boverall, Prompt· Berkeley Function-Calling Leaderboard data40.4%
34GPT 4.1overall, Prompt· Berkeley Function-Calling Leaderboard data39.4%
35Mistral Largeoverall, fc· Berkeley Function-Calling Leaderboard data38.4%
36Qwen3 14Boverall, Prompt· Berkeley Function-Calling Leaderboard data37.8%
37Mistral Medium 3overall· Berkeley Function-Calling Leaderboard data37.7%
38Mistral Medium 3overall, fc· Berkeley Function-Calling Leaderboard data37.6%
39Llama 4 Maverickoverall, fc· Berkeley Function-Calling Leaderboard data37.3%
40Mistral Smalloverall, fc· Berkeley Function-Calling Leaderboard data37.2%
41Gemini 2.5 Flash-Liteoverall, fc· Berkeley Function-Calling Leaderboard data36.9%
42Qwen3 30B A3B Instructoverall, Prompt· Berkeley Function-Calling Leaderboard data36.7%
43Qwen3 4B Instruct 2507overall, fc· Berkeley Function-Calling Leaderboard data35.7%
44Qwen3 4B Instruct 2507overall, Prompt· Berkeley Function-Calling Leaderboard data35.5%
45Claude Opus 4.5overall, Prompt· Berkeley Function-Calling Leaderboard data33.5%
46GPT-4.1 nanooverall, fc· Berkeley Function-Calling Leaderboard data33%
47Mistral Smalloverall, Prompt· Berkeley Function-Calling Leaderboard data32.4%
48command-r7boverall, fc· Berkeley Function-Calling Leaderboard data32.1%
49Llama 3.3 70B Instructoverall, fc· Berkeley Function-Calling Leaderboard data31.9%
50Mistral Largeoverall, Prompt· Berkeley Function-Calling Leaderboard data31.8%

Showing 1–50 of 76 results.

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