Skip to content

Benchmark

BFCL Multi-Turn

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

Models measured
50
Scores recorded
78
Sources
2
IQ weight
1.04

Score distribution

25.3%
Mean
20.3%
Median
21.5%
Std dev
1.6%
10th percentile
58.9%
90th percentile
0%
Min
0%76.8%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.04 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 27B ITPrompt· Berkeley Function-Calling Leaderboard data10.8%
52Mistral Medium 3fc· Berkeley Function-Calling Leaderboard data10.8%
53Mistral Medium 3· Berkeley Function-Calling Leaderboard data9.9%
54GPT 4.1Prompt· Berkeley Function-Calling Leaderboard data9.8%
55Llama 4 Scoutfc· Berkeley Function-Calling Leaderboard data9%
56command-r7bfc· Berkeley Function-Calling Leaderboard data8.3%
57Open Mistral Nemofc· Berkeley Function-Calling Leaderboard data7.8%
58Gemini 2.5 Flash-LitePrompt· Berkeley Function-Calling Leaderboard data7.6%
59Granite 3.1 8B Instructfc· Berkeley Function-Calling Leaderboard data7.5%
60Gemma 3 12B ITPrompt· Berkeley Function-Calling Leaderboard data5.8%
61GPT-5 miniPrompt· Berkeley Function-Calling Leaderboard data5.5%
62Llama 3.2 3Bfc· Berkeley Function-Calling Leaderboard data4%
63Phi 4Prompt· Berkeley Function-Calling Leaderboard data3.9%
64Qwen3 0.6Bfc· Berkeley Function-Calling Leaderboard data3.6%
65GPT-4.1 miniPrompt· Berkeley Function-Calling Leaderboard data2.5%
66Nova 2 Litefc· Berkeley Function-Calling Leaderboard data2.1%
67GPT-4.1 nanoPrompt· Berkeley Function-Calling Leaderboard data2%
68Nova Profc· Berkeley Function-Calling Leaderboard data1.9%
69Claude Haiku 4.5Prompt· Berkeley Function-Calling Leaderboard data1.8%
70Claude Sonnet 4.5Prompt· Berkeley Function-Calling Leaderboard data1.6%
71Nova Microfc· Berkeley Function-Calling Leaderboard data1.4%
72Qwen3 0.6BPrompt· Berkeley Function-Calling Leaderboard data1.4%
73GPT-5 nanoPrompt· Berkeley Function-Calling Leaderboard data0.75%
74Open Mistral NemoPrompt· Berkeley Function-Calling Leaderboard data0.75%
75Gemma 3 4B ITPrompt· Berkeley Function-Calling Leaderboard data0.38%
76Gemma 3 1b itPrompt· Berkeley Function-Calling Leaderboard data0%
77Llama 3.1 Nemotron Ultra 253B v1fc· Berkeley Function-Calling Leaderboard data0%
78Llama 3.2 1B Instructfc· Berkeley Function-Calling Leaderboard data0%

Showing 51–78 of 78 results.

Data credits: Berkeley Function-Calling Leaderboard (Gorilla, UC Berkeley), Apache-2.0 (licence) · BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.