Skip to content

Benchmark

BFCL Web Search

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
1.16

Score distribution

32.4%
Mean
21%
Median
29.9%
Std dev
1%
10th percentile
77%
90th percentile
0%
Min
0%84.5%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.16 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 4.5fc· Berkeley Function-Calling Leaderboard data84.5%
2Claude Haiku 4.5fc· Berkeley Function-Calling Leaderboard data83.5%
3Grok 4.1 Fastfc· Berkeley Function-Calling Leaderboard data82.5%
4GPT-5 minifc· Berkeley Function-Calling Leaderboard data82%
5Grok 4fc· Berkeley Function-Calling Leaderboard data82%
6Claude Sonnet 4.5fc· Berkeley Function-Calling Leaderboard data81%
7Gemini 3 Pro PreviewPrompt· Berkeley Function-Calling Leaderboard data80%
8GLM 4.6FC thinking· Berkeley Function-Calling Leaderboard data77.5%
9OpenAI o3fc· Berkeley Function-Calling Leaderboard data77%
10GPT 5.2fc· Berkeley Function-Calling Leaderboard data75.5%
11o4 Minifc· Berkeley Function-Calling Leaderboard data75.5%
12Grok 4Prompt· Berkeley Function-Calling Leaderboard data74%
13GPT-5 nanofc· Berkeley Function-Calling Leaderboard data72.5%
14o4 MiniPrompt· Berkeley Function-Calling Leaderboard data71.5%
15DeepSeek V3.2 Expfc· Berkeley Function-Calling Leaderboard data69.5%
16Gemini 3 Pro Previewfc· Berkeley Function-Calling Leaderboard data68.5%
17GPT 4.1fc· Berkeley Function-Calling Leaderboard data68%
18Kimi K2 Instructfc· Berkeley Function-Calling Leaderboard data66.5%
19Gemini 2.5 FlashPrompt· Berkeley Function-Calling Leaderboard data62%
20Gemini 2.5 Flashfc· Berkeley Function-Calling Leaderboard data59%
21DeepSeek V3.2 ExpPrompt + Thinking· Berkeley Function-Calling Leaderboard data58%
22GPT-4.1 minifc· Berkeley Function-Calling Leaderboard data57%
23Command Afc· Berkeley Function-Calling Leaderboard data55.5%
24Qwen3 235B A22B Instruct 2507fc· Berkeley Function-Calling Leaderboard data54%
25OpenAI o3Prompt· Berkeley Function-Calling Leaderboard data50.5%
26Qwen3 235B A22B Instruct 2507Prompt· Berkeley Function-Calling Leaderboard data50.5%
27GPT 5.2Prompt· Berkeley Function-Calling Leaderboard data40.5%
28Mistral Medium 3· Berkeley Function-Calling Leaderboard data39%
29GPT 4.1Prompt· Berkeley Function-Calling Leaderboard data35%
30Mistral Medium 3fc· Berkeley Function-Calling Leaderboard data35%
31Mistral Smallfc· Berkeley Function-Calling Leaderboard data31%
32Llama 4 Maverickfc· Berkeley Function-Calling Leaderboard data28%
33Mistral Largefc· Berkeley Function-Calling Leaderboard data28%
34command-r7bfc· Berkeley Function-Calling Leaderboard data27%
35Qwen3 32BPrompt· Berkeley Function-Calling Leaderboard data26%
36Qwen3 30B A3B Instructfc· Berkeley Function-Calling Leaderboard data22.5%
37Qwen3 32Bfc· Berkeley Function-Calling Leaderboard data21.5%
38Gemini 2.5 Flash-Litefc· Berkeley Function-Calling Leaderboard data21%
39Mistral LargePrompt· Berkeley Function-Calling Leaderboard data20%
40Claude Haiku 4.5Prompt· Berkeley Function-Calling Leaderboard data19.5%
41Qwen3 30B A3B InstructPrompt· Berkeley Function-Calling Leaderboard data17.5%
42Claude Sonnet 4.5Prompt· Berkeley Function-Calling Leaderboard data16%
43Llama 4 Scoutfc· Berkeley Function-Calling Leaderboard data14.5%
44GPT-5 nanoPrompt· Berkeley Function-Calling Leaderboard data13.5%
45Qwen3 8BPrompt· Berkeley Function-Calling Leaderboard data13.5%
46Claude Opus 4.5Prompt· Berkeley Function-Calling Leaderboard data13%
47Qwen3 8Bfc· Berkeley Function-Calling Leaderboard data12%
48GPT-4.1 nanofc· Berkeley Function-Calling Leaderboard data11%
49Qwen3 14BPrompt· Berkeley Function-Calling Leaderboard data10.5%
50Llama 3.3 70B Instructfc· Berkeley Function-Calling Leaderboard data10%

Showing 1–50 of 76 results.

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