Berkeley Function Calling Leaderboard (BFCL) V4

Name: Berkeley Function Calling Leaderboard (BFCL) V4 results
Creator: Berkeley Function Calling Leaderboard (BFCL) V4
License: https://github.com/ShishirPatil/gorilla/blob/main/LICENSE

BFCL's own weighted average across its test categories.

Data older than 90 days: last updated 16 Dec 2025

Results dated: 16 Dec 2025
Results: 109 configurations of 81 models
Unit: % correct
Licence: Apache License 2.0

Overall accuracy: Grok 4 (0709)

Top 15 of 81 results · % correct, higher is better. Choose a model to highlight it.Clear highlight

1 Claude Opus 4.5Function callingAnthropic 77.5%
2 Claude Sonnet 4.5Function callingAnthropic 73.2%
3 Gemini 3 Pro PreviewPromptingGoogle 72.5%
4 GLM-4.6 (reasoning on)Function callingZ.ai 72.4%
5 Grok 4.1 Fast (reasoning)Function callingxAI 69.6%
6 Claude Haiku 4.5Function callingAnthropic 68.7%
7 o3PromptingOpenAI 63.1%
8 Grok 4 (0709)PromptingxAI 63.0%
9 Kimi K2 (0711)Function callingMoonshot AI 59.1%
10 Grok 4.1 Fast (non-reasoning)Function callingxAI 58.3%
11 Command A ReasoningFunction callingCohere 57.1%
12 DeepSeek-V3.2-Exp (reasoning on)PromptingDeepSeek 56.7%
13 Gemini 2.5 FlashFunction callingGoogle 56.2%
14 GPT-5.2Function callingOpenAI 55.9%
15 GPT-5 miniFunction callingOpenAI 55.5%

Full results

BFCL: overall accuracy, % correct, higher is better
#	Model	Overall accuracy % correct, higher is better	Price $ per million tokens, in / out
1	Claude Opus 4.5Function callingAnthropic · best of 2 settings	77.5%	$5 / $25
2	Claude Sonnet 4.5Function callingAnthropic · best of 2 settings	73.2%	$3 / $15
3	Gemini 3 Pro PreviewPromptingGoogle · best of 2 settings	72.5%	–
4	GLM-4.6 (reasoning on)Function callingZ.ai	72.4%	$0.50 / $2
5	Grok 4.1 Fast (reasoning)Function callingxAI	69.6%	–
6	Claude Haiku 4.5Function callingAnthropic · best of 2 settings	68.7%	$1 / $5
7	o3PromptingOpenAI · best of 2 settings	63.1%	$2 / $8
8	Grok 4 (0709)PromptingxAI · best of 2 settings	63.0%	–
9	Kimi K2 (0711)Function callingMoonshot AI	59.1%	$0.57 / $2.30
10	Grok 4.1 Fast (non-reasoning)Function callingxAI	58.3%	–
11	Command A ReasoningFunction callingCohere	57.1%	–
12	DeepSeek-V3.2-Exp (reasoning on)PromptingDeepSeek · best of 2 settings	56.7%	$0.27 / $0.41
13	Gemini 2.5 FlashFunction callingGoogle · best of 2 settings	56.2%	$0.30 / $2.50
14	GPT-5.2Function callingOpenAI · best of 2 settings	55.9%	$1.75 / $14
15	GPT-5 miniFunction callingOpenAI · best of 2 settings	55.5%	$0.25 / $2
16	xLAM-2-32b-fc-rFunction callingSalesforce	54.7%	–
17	GPT-4.1Function callingOpenAI · best of 2 settings	54.0%	$2 / $8
18	o4-miniFunction callingOpenAI · best of 2 settings	53.2%	$1.10 / $4.40
19	xLAM-2-70b-fc-rFunction callingSalesforce	53.1%	–
20	Qwen3-235B-A22B-Instruct-2507PromptingAlibaba · best of 2 settings	52.2%	$0.15 / $0.75
21	GPT-5 nanoFunction callingOpenAI · best of 2 settings	51.5%	$0.05 / $0.40
22	Nanbeige4-3B-Thinking (2511)Function callingNanbeige	51.4%	–
23	GPT-4.1 miniFunction callingOpenAI · best of 2 settings	50.5%	$0.40 / $1.60
24	Qwen3-32BFunction callingAlibaba · best of 2 settings	48.7%	$0.14 / $0.40
25	Nanbeige3.5-Pro-ThinkingFunction callingNanbeige	47.7%	–
26	xLAM-2-8b-fc-rFunction callingSalesforce	46.7%	–
27	Command AFunction callingCohere	46.5%	$2.50 / $10
28	BitAgent Bounty 8BMode not statedBittensor	46.2%	–
29	Arch-Agent-32BMode not statedKatanemo	45.4%	–
30	Qwen3-8BFunction callingAlibaba · best of 2 settings	42.6%	$0.12 / $0.46
31	ToolACE-2-8BFunction callingHuawei Noah's Ark Lab and USTC	42.4%	–
32	Qwen3-30B-A3B-Instruct-2507Function callingAlibaba · best of 2 settings	41.4%	$0.09 / $0.30
33	xLAM-2-3b-fc-rFunction callingSalesforce	41.2%	–
34	Qwen3-14BFunction callingAlibaba · best of 2 settings	41.0%	$0.12 / $0.24
35	Mistral Large 2.1 (2411)Function callingMistral AI · best of 2 settings	38.4%	–
36	Mistral Medium 3Mode not statedMistral AI · best of 2 settings	37.7%	$0.40 / $2
37	Llama 4 Maverick 17B 128E Instruct (FP8)Function callingMeta	37.3%	–
38	Mistral Small 3.2 24BFunction callingMistral AI · best of 2 settings	37.2%	$0.094 / $0.25
39	Gemini 2.5 Flash-LiteFunction callingGoogle · best of 2 settings	36.9%	$0.10 / $0.40
40	Qwen3-4B-Instruct-2507Function callingAlibaba · best of 2 settings	35.7%	–
41	Arch-Agent-3BMode not statedKatanemo	35.4%	–
42	GPT-4.1 nanoFunction callingOpenAI · best of 2 settings	33.1%	$0.10 / $0.40
43	Arch-Agent-1.5BMode not statedKatanemo	32.1%	–
43	Command R7B (12-2024)Function callingCohere	32.1%	$0.037 / $0.15
45	Llama 3.3 70B InstructFunction callingMeta	31.9%	$0.59 / $0.79
46	Hammer2.1-7BFunction callingMadeAgents	31.7%	–
47	xLAM-2-1b-fc-rFunction callingSalesforce	30.4%	–
47	Gemma 3 12BPromptingGoogle	30.4%	$0.05 / $0.15
49	Hammer2.1-3BFunction callingMadeAgents	29.7%	–
50	Gemma 3 27BPromptingGoogle	29.5%	$0.12 / $0.20
51	Phi-4PromptingMicrosoft	28.8%	$0.07 / $0.14
52	Qwen3-1.7BFunction callingAlibaba	28.4%	–
53	Llama 4 ScoutFunction callingMeta	28.1%	$0.18 / $0.59
54	CoALM 70BMode not statedUIUC and Oumi	28.0%	–
55	Hammer2.1-1.5BFunction callingMadeAgents	27.9%	–
55	Palmyra X 004Function callingWriter	27.9%	–
57	Mistral NemoFunction callingMistral AI · best of 2 settings	27.6%	$0.023 / $0.03
58	Nova 2 LiteFunction callingAmazon	27.1%	$0.30 / $2.50
58	Granite 3.1 8B InstructFunction callingIBM	27.1%	–
60	Falcon3 10B InstructFunction callingTII	27.0%	–
61	Granite 3.2 8B InstructFunction callingIBM	26.9%	–
62	CoALM 8BMode not statedUIUC and Oumi	26.8%	–
63	Llama 3.1 8B InstructPromptingMeta	25.8%	$0.05 / $0.08
64	MiniCPM3-4BFunction callingOpenBMB · best of 2 settings	25.6%	–
65	Nova Pro 1.0Function callingAmazon	25.0%	$0.80 / $3.20
66	Falcon3 7B InstructFunction callingTII	24.0%	–
67	Qwen3-0.6BFunction callingAlibaba · best of 2 settings	23.9%	–
68	Granite 20B FunctionCallingFunction callingIBM	23.2%	–
69	Nova Micro 1.0Function callingAmazon	22.3%	$0.035 / $0.14
69	RZN-TPromptingPhronetic AI	22.3%	–
71	Llama 3.2 3B InstructFunction callingMeta	22.0%	$0.05 / $0.33
72	Bielik 11B v2.3 InstructPromptingSpeakLeash and ACK Cyfronet AGH	21.9%	–
73	Hammer2.1-0.5BFunction callingMadeAgents	21.2%	–
74	Gemma 3 4BPromptingGoogle	19.6%	$0.05 / $0.10
75	Granite 4.0 350MFunction callingIBM	19.0%	–
76	Falcon3 3B InstructFunction callingTII	16.3%	–
77	Ministral 8B (2410)Function callingMistral AI	11.1%	–
77	Falcon3 1B InstructFunction callingTII	11.1%	–
79	Llama 3.2 1B InstructFunction callingMeta	10.8%	$0.027 / $0.20
80	Llama 3.1 Nemotron Ultra 253B v1Function callingNVIDIA	10.0%	–
81	Gemma 3 1BPromptingGoogle	7.2%	–

Swipe the table sideways for more columns.

Ranks follow the score as shown, so equal numbers share a rank. Each model is shown at its best setting; show every setting. Results as published by Berkeley Function Calling Leaderboard (BFCL) V4; we do not re-run them.

Progress over time

Overall accuracy, plotted at each model's release date, higher is better. Each model counts once, at its best setting. Scores are today's, not launch day's; 38 models without a known release date are left off. Hover or tap a mark for its model.

Best score so far

Every model at its release date. The line moves each time a new release beats every earlier one, and each new best is named where there is room.

Best score by lab

Each lab's best so far, stepping whenever one of its releases beats its own earlier models. The 6 labs with the best scores today; 7 others not drawn.

AnthropicGoogleZ.aixAIOpenAIMoonshot AI

What it measures

BFCL's own weighted average across its test categories.

What it does not measure

Not a neutral average: the weighting is BFCL's. Not reliability on your own tools: BFCL's functions and queries are a fixed test set, and a correct call is judged by its form, not by what it achieved.

Source

Berkeley Function Calling Leaderboard (BFCL) V4 · Methodology

Berkeley Function Calling Leaderboard (BFCL) V4 by the UC Berkeley Gorilla team, Apache License 2.0.