allbenched allbenched Allbenched allbenched allbenched AllBenched allbenched allbenched allbenched

Benchmarks / Agents

BFCL v4

Function and tool calling: single, parallel and multi-turn calls, web search and memory agent tasks, and knowing when not to call a tool. Run by the Berkeley Gorilla team; FC means native function calling, Prompt means tools described in the prompt.

CategoryAgents
MetricAccuracy (%)
SourceBFCL (original)
LicenseApache 2.0
Benchmark pagegorilla.cs.berkeley.edu
StatusScored
Weight0.8 within Agents
FitDiscrimination 0.51, noise 0.19

Results

#ModelProviderResultDate
1 Claude Opus 4.5 Anthropic 77.5% 2026-04-12
2 Claude Sonnet 4.5 Anthropic 73.2% 2026-04-12
3 Gemini 3 Pro Preview Google 72.5% 2026-04-12
4 GLM-4.6 Zhipu AI 72.4% 2026-04-12
5 Grok 4.1 Fast (Reasoning) xAI 69.6% 2026-04-12
6 Claude Haiku 4.5 Anthropic 68.7% 2026-04-12
7 Gemini 3 Pro Preview Google 68.1% 2026-04-12
8 o3 OpenAI 63.1% 2026-04-12
9 Grok 4 xAI 63.0% 2026-04-12
10 Grok 4 xAI 61.4% 2026-04-12
11 Kimi K2 Instruct Moonshot AI 59.1% 2026-04-12
12 Grok 4.1 Fast xAI 58.3% 2026-04-12
13 Command A Reasoning Cohere 57.1% 2026-04-12
14 DeepSeek-V3.2-Exp DeepSeek 56.7% 2026-04-12
15 Gemini 2.5 Flash Google 56.2% 2026-04-12
16 GPT-5.2 OpenAI 55.9% 2026-04-12
17 GPT-5 Mini OpenAI 55.5% 2026-04-12
18 DeepSeek-V3.2-Exp DeepSeek 54.1% 2026-04-12
19 GPT-4.1 OpenAI 54.0% 2026-04-12
20 o4-mini OpenAI 53.2% 2026-04-12
21 Qwen3 235B-A22B Instruct 2507 Alibaba 52.2% 2026-04-12
22 GPT-5 Nano OpenAI 51.5% 2026-04-12
23 Gemini 2.5 Flash Google 50.9% 2026-04-12
24 GPT-4.1 mini OpenAI 50.5% 2026-04-12
25 o4-mini OpenAI 50.3% 2026-04-12
26 Qwen3 32B Alibaba 48.7% 2026-04-12
27 o3 OpenAI 48.6% 2026-04-12
28 Qwen3 235B-A22B Instruct 2507 Alibaba 48.0% 2026-04-12
29 Qwen3 32B Alibaba 46.8% 2026-04-12
30 Command A Cohere 46.5% 2026-04-12
31 GPT-5.2 OpenAI 45.3% 2026-04-12
32 Qwen3 8B Alibaba 42.6% 2026-04-12
33 Qwen3 30B-A3B Instruct 2507 Alibaba 41.4% 2026-04-12
34 Qwen3-14B Alibaba 41.0% 2026-04-12
35 Qwen3 8B Alibaba 40.4% 2026-04-12
36 GPT-4.1 OpenAI 39.4% 2026-04-12
37 Mistral Large 2.1 Mistral 38.4% 2026-04-12
38 Qwen3-14B Alibaba 37.8% 2026-04-12
39 Mistral Medium 3 Mistral 37.7% 2026-04-12
40 Mistral Medium 3 Mistral 37.6% 2026-04-12
41 Llama 4 Maverick Meta 37.3% 2026-04-12
42 Mistral Small 3.2 Mistral 37.2% 2026-04-12
43 Gemini 2.5 Flash-Lite Google 36.9% 2026-04-12
44 Qwen3 30B-A3B Instruct 2507 Alibaba 36.7% 2026-04-12
45 qwen3-4b-instruct-2507 Unknown 35.7% 2026-04-12
46 qwen3-4b-instruct-2507 Unknown 35.5% 2026-04-12
47 Claude Opus 4.5 Anthropic 33.5% 2026-04-12
48 GPT-4.1 nano OpenAI 33.1% 2026-04-12
49 Mistral Small 3.2 Mistral 32.4% 2026-04-12
50 Command R7B Cohere 32.1% 2026-04-12
51 Llama-3.3-70B-Instruct Meta 31.9% 2026-04-12
52 Mistral Large 2.1 Mistral 31.8% 2026-04-12
53 Gemma 3 12B IT Google 30.4% 2026-04-12
54 GPT-4.1 mini OpenAI 29.7% 2026-04-12
55 Gemma 3 27B IT Google 29.5% 2026-04-12
56 Phi-4 Microsoft 28.8% 2026-04-12
57 Qwen3-1.7B Alibaba 28.4% 2026-04-12
58 Llama 4 Scout Meta 28.1% 2026-04-12
59 Gemini 2.5 Flash-Lite Google 28.0% 2026-04-12
60 Palmyra X4 Writer 27.9% 2026-04-12
61 GPT-5 Mini OpenAI 27.8% 2026-04-12
62 Mistral Nemo Mistral 27.6% 2026-04-12
63 GPT-5 Nano OpenAI 27.6% 2026-04-12
64 Nova 2 Lite Amazon 27.1% 2026-04-12
65 Llama-3.1-8B-Instruct Meta 25.8% 2026-04-12
66 Claude Haiku 4.5 Anthropic 25.3% 2026-04-12
67 Amazon Nova Pro Amazon 25.0% 2026-04-12
68 Claude Sonnet 4.5 Anthropic 24.9% 2026-04-12
69 GPT-4.1 nano OpenAI 24.9% 2026-04-12
70 Qwen3-0.6B Alibaba 23.9% 2026-04-12
71 Qwen3-0.6B Alibaba 22.4% 2026-04-12
72 Amazon Nova Micro Amazon 22.3% 2026-04-12
73 Llama 3.2 3B Instruct Meta 22.0% 2026-04-12
74 Gemma 3 4B IT Google 19.6% 2026-04-12
75 Mistral Nemo Mistral 19.3% 2026-04-12
76 granite-4.0-350m Unknown 19.0% 2026-04-12
77 Ministral 8B Instruct Mistral 11.1% 2026-04-12
78 Llama 3.2 1B Instruct Meta 10.8% 2026-04-12
79 Llama 3.1 Nemotron Ultra 253B NVIDIA 10.0% 2026-04-12
80 Gemma 3 1B Google 7.2% 2026-04-12