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
1MiniMax M2.5main capability figure· Reported by MiniMax· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.76.8%
2MiniMax M2.5search/tool figure· Reported by MiniMax· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.76.8%
3Claude Opus 4.5fc· Berkeley Function-Calling Leaderboard data68.4%
4GLM 4.6FC thinking· Berkeley Function-Calling Leaderboard data68%
5Gemini 3 Pro Previewfc· Berkeley Function-Calling Leaderboard data63.1%
6OpenAI o3Prompt· Berkeley Function-Calling Leaderboard data62.3%
7Claude Sonnet 4.5fc· Berkeley Function-Calling Leaderboard data61.4%
8Gemini 3 Pro PreviewPrompt· Berkeley Function-Calling Leaderboard data60.8%
9Grok 4.1 Fastfc· Berkeley Function-Calling Leaderboard data58.9%
10Claude Haiku 4.5fc· Berkeley Function-Calling Leaderboard data53.6%
11Kimi K2 Instructfc· Berkeley Function-Calling Leaderboard data50.6%
12Command Afc· Berkeley Function-Calling Leaderboard data50.1%
13Qwen3 32Bfc· Berkeley Function-Calling Leaderboard data47.9%
14Grok 4Prompt· Berkeley Function-Calling Leaderboard data47%
15Qwen3 235B A22B Instruct 2507fc· Berkeley Function-Calling Leaderboard data45.4%
16DeepSeek V3.2 ExpPrompt + Thinking· Berkeley Function-Calling Leaderboard data44.9%
17Qwen3 235B A22B Instruct 2507Prompt· Berkeley Function-Calling Leaderboard data44.6%
18GPT 5.2Prompt· Berkeley Function-Calling Leaderboard data43.8%
19Qwen3 32BPrompt· Berkeley Function-Calling Leaderboard data43.3%
20o4 Minifc· Berkeley Function-Calling Leaderboard data41.8%
21Qwen3 8Bfc· Berkeley Function-Calling Leaderboard data41.8%
22GPT 4.1fc· Berkeley Function-Calling Leaderboard data38.9%
23DeepSeek V3.2 Expfc· Berkeley Function-Calling Leaderboard data37.4%
24Gemini 2.5 Flashfc· Berkeley Function-Calling Leaderboard data36.3%
25Qwen3 14Bfc· Berkeley Function-Calling Leaderboard data34.8%
26GPT-5 nanofc· Berkeley Function-Calling Leaderboard data34.5%
27GPT-4.1 minifc· Berkeley Function-Calling Leaderboard data34.1%
28Grok 4fc· Berkeley Function-Calling Leaderboard data33.9%
29Qwen3 8BPrompt· Berkeley Function-Calling Leaderboard data33.4%
30Qwen3 30B A3B Instructfc· Berkeley Function-Calling Leaderboard data30%
31GPT 5.2fc· Berkeley Function-Calling Leaderboard data28.1%
32GPT-5 minifc· Berkeley Function-Calling Leaderboard data27.5%
33Qwen3 14BPrompt· Berkeley Function-Calling Leaderboard data26.1%
34GPT-4.1 nanofc· Berkeley Function-Calling Leaderboard data23.6%
35Qwen3 30B A3B InstructPrompt· Berkeley Function-Calling Leaderboard data23.5%
36Qwen3 4B Instruct 2507fc· Berkeley Function-Calling Leaderboard data22.1%
37Llama 3.3 70B Instructfc· Berkeley Function-Calling Leaderboard data21.5%
38Qwen3 4B Instruct 2507Prompt· Berkeley Function-Calling Leaderboard data20.5%
39Llama 4 Maverickfc· Berkeley Function-Calling Leaderboard data20.3%
40Gemini 2.5 FlashPrompt· Berkeley Function-Calling Leaderboard data16.8%
41o4 MiniPrompt· Berkeley Function-Calling Leaderboard data16.6%
42Claude Opus 4.5Prompt· Berkeley Function-Calling Leaderboard data16.1%
43Mistral SmallPrompt· Berkeley Function-Calling Leaderboard data14.8%
44OpenAI o3fc· Berkeley Function-Calling Leaderboard data14.8%
45Mistral Largefc· Berkeley Function-Calling Leaderboard data14.1%
46Mistral LargePrompt· Berkeley Function-Calling Leaderboard data13.8%
47Gemini 2.5 Flash-Litefc· Berkeley Function-Calling Leaderboard data13.5%
48Mistral Smallfc· Berkeley Function-Calling Leaderboard data11.5%
49Llama 3.1 8B InstructPrompt· Berkeley Function-Calling Leaderboard data11.1%
50Qwen3 1.7Bfc· Berkeley Function-Calling Leaderboard data11%

Showing 1–50 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.