BFCL

Agentic

Berkeley Function Calling Leaderboard — function/tool calling accuracy

Source ↗ Higher is better

Rank	Model	Provider	Score
1	Anthropic: Claude Opus 4.5	Anthropic	79.1%
2	Anthropic: Claude Sonnet 4.5	Anthropic	73.0%
3	Gemini 3 Pro Preview	Google	70.9%
4	Grok 4	SpaceXAI	69.0%
5	Anthropic: Claude Haiku 4.5	Anthropic	69.0%
6	grok-4-1-fast-reasoning	SpaceXAI	68.2%
7	GLM 4.6 (FC thinking)	Z.ai	66.6%
8	GPT 5 mini 2025 08 07	OpenAI	63.2%
9	o3-2025-04-16	OpenAI	62.2%
10	DeepSeek: DeepSeek V3.2 ExpHistorical	DeepSeek	61.9%
13	↳DeepSeek V3.2 Exp (Prompt + Thinking)	DeepSeek	51.0%
11	GPT 5.2 2025 12 11	OpenAI	60.7%
12	o4-mini-2025-04-16	OpenAI	54.9%
14	Grok 4.1 Fast (Non-reasoning)	SpaceXAI	50.6%
15	Google: Gemini 2.5 FlashRetiring	Google	50.4%
16	GPT 5 nano 2025 08 07	OpenAI	48.6%
17	Moonshotai Kimi K2 Instruct	Moonshot AI	47.8%
18	gpt-4.1-2025-04-14	OpenAI	45.9%
19	Nanbeige3.5 Pro Thinking	Nanbeige	43.6%
20	Command A Reasoning	Cohere	42.2%
21	gpt-4.1-mini-2025-04-14	OpenAI	41.9%
22	qwen3-235b-a22b-instruct-2507	Alibaba	38.9%
23	Cohere: Command A	Cohere	31.5%
24	mistral-medium-2505	Mistral AI	30.4%
25	Nanbeige4 3B Thinking 2511	Nanbeige	29.1%
26	mistral-large-2411	Mistral AI	26.5%
27	mistral-small-2506	Mistral AI	24.5%
28	Qwen: Qwen3 32B	Alibaba	24.1%
29	Llama 4 Maverick 17B 128E Instruct FP8	Meta	23.5%
30	xLAM 2 32b fc r	Salesforce	23.2%
31	Google: Gemini 2.5 Flash LiteRetiring	Google	20.8%
32	Qwen: Qwen3 30B A3B Instruct 2507	Alibaba	20.1%
33	Command R7B	Cohere	16.1%
34	Google: Gemma 3 12B	Google	15.8%
35	gpt-4.1-nano-2025-04-14	OpenAI	15.0%
36	Qwen: Qwen3 14B	Alibaba	14.8%
37	xLAM 2 70b fc r	Salesforce	14.7%
38	Microsoft: Phi 4	Microsoft	14.6%
39	tiiuae/Falcon3-10B-Instruct	TII UAE	14.5%
40	Qwen3 4B Instruct 2507	Alibaba	14.2%
41	ToolACE 2 8B	Huawei	13.5%
42	Qwen: Qwen3 8BHistorical	Alibaba	13.3%
43	llama-4-scout-17b-16e-instruct	Meta	11.3%
44	tiiuae/Falcon3-7B-Instruct	TII UAE	10.6%
45	xLAM 2 8b fc r	Salesforce	10.2%
46	Arch Agent 32B	Katanemo	9.8%
47	Meta: Llama 3.3 70B Instruct	Meta	9.1%
48	Open Mistral Nemo 2407	Mistral AI	8.7%
49	palmyra x 004	Writer	7.8%
50	ibm-granite/granite-3.1-8b-instruct	IBM	7.5%

Access this data

JSON GET /api/v1

[
  {
    "rank": 1,
    "model": "Anthropic: Claude Opus 4.5",
    "provider": "Anthropic",
    "score": "79.13"
  },
  {
    "rank": 2,
    "model": "Anthropic: Claude Sonnet 4.5",
    "provider": "Anthropic",
    "score": "72.97"
  },
  {
    "rank": 3,
    "model": "Gemini 3 Pro Preview",
    "provider": "Google",
    "score": "70.86"
  }
]

Curl

$ curl -s https://rightrank.ai/api/v1/rankings/bfcl.json | jq '.[:3]'

CLI brew install rightrank

$ rightrank top bfcl --limit 5
$ rightrank compare claude-opus-4.6 gpt-5.4-pro --category bfcl
$ rightrank value bfcl  # best performance per dollar

Claude Code Skill Skill

# In Claude Code, use the RightRank skill:
/rightrank best model for bfcl
/rightrank compare top 5 by value
/rightrank cheapest model scoring above 80 on bfcl