Agentic · tools

BFCL v4 leaderboard

Name: benchgap: LLM benchmark scores, measured and estimated
Creator: benchgap

As of 2026-10-07, the highest measured score on BFCL v4 is 88.5% by BTL-3. 187 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	K-EXAONE 2.0	100.0%	estimated ± 8.6 pp, low confidence
2	Ornith-1.0-397B	100.0%	estimated ± 8.6 pp, low confidence
3	Ornith-1.5-397B	100.0%	estimated ± 8.6 pp, low confidence
4	dots3-note Preview	93.7%	estimated ± 8.6 pp, low confidence
5	Ornith-1.5-35B-A3B	91.2%	estimated ± 8.6 pp, low confidence
6	BTL-3	88.5%	measured
7	Claude Opus 4.6	85.2%	estimated ± 8.6 pp, low confidence
8	Ornith-1.0-35B	83.5%	estimated ± 8.6 pp, low confidence
9	Claude Opus 5.5	78.6%	estimated ± 2.6 pp, medium confidence
10	Claude Sonnet 5.5	78.5%	estimated ± 2.6 pp, medium confidence
11	Claude Fable 5.1	78.3%	estimated ± 2.6 pp, medium confidence
12	Claude Opus 5	78.2%	estimated ± 2.6 pp, medium confidence
13	Grok 4.7	78.2%	estimated ± 2.6 pp, medium confidence
14	MiMo-V2.6-Pro	78.1%	estimated ± 2.6 pp, medium confidence
15	Muse Spark 1.3	78.1%	estimated ± 2.6 pp, medium confidence
16	Qwen3.8 Max Preview	78.1%	estimated ± 2.6 pp, medium confidence
17	GLM-5.3	78.0%	estimated ± 2.6 pp, medium confidence
18	Qwen3.8-Flash-Next	77.9%	estimated ± 2.6 pp, medium confidence
19	Gemini 4 Argon	77.9%	estimated ± 2.6 pp, medium confidence
20	Grok 4.6	77.9%	estimated ± 2.6 pp, medium confidence
21	Ling 3.1 Flash	77.9%	estimated ± 2.6 pp, medium confidence
22	Claude Fable 5	77.9%	estimated ± 2.6 pp, medium confidence
23	GPT-5.6 Sol	77.9%	estimated ± 2.6 pp, medium confidence
24	MiMo-V2.6-Flash	77.9%	estimated ± 2.6 pp, medium confidence
25	Claude Sonnet 4.6	77.9%	estimated ± 8.6 pp, low confidence
26	DeepSeek V4.1 Flash	77.8%	estimated ± 2.6 pp, medium confidence
27	DeepSeek V4 Pro 0813	77.8%	estimated ± 2.6 pp, medium confidence
28	Step 5 Preview	77.8%	estimated ± 2.6 pp, medium confidence
29	GPT-6.1 Sol	77.8%	estimated ± 2.6 pp, medium confidence
30	GPT-6 Astra	77.6%	estimated ± 2.6 pp, medium confidence
31	Kimi K3	77.6%	estimated ± 2.6 pp, medium confidence
32	GPT-6 Sol	77.5%	estimated ± 2.6 pp, medium confidence
33	Muse Spark 1.2	77.4%	estimated ± 2.6 pp, medium confidence
34	Claude Sonnet 5	77.3%	estimated ± 2.6 pp, medium confidence
35	GPT-5.6 Luna	77.3%	estimated ± 2.6 pp, medium confidence
36	Claude Opus 4.8	77.3%	estimated ± 2.6 pp, medium confidence
37	GPT-5.6 Terra	77.3%	estimated ± 2.6 pp, medium confidence
38	DeepSeek V4 Flash 0731	77.2%	estimated ± 2.6 pp, medium confidence
39	GPT-6 Luna	77.2%	estimated ± 2.6 pp, medium confidence
40	Gemini 3.8 Flash	77.2%	estimated ± 2.6 pp, medium confidence
41	Mistral Large 4	77.1%	estimated ± 2.6 pp, medium confidence
42	Qwen3.8-27B	77.1%	estimated ± 2.6 pp, medium confidence
43	Gemini 3.7 Flash	77.0%	estimated ± 2.6 pp, medium confidence
44	Atria Dawn Preview	77.0%	measured
45	Grok 4.5	77.0%	estimated ± 2.6 pp, medium confidence
46	GLM-5.2	76.9%	estimated ± 2.6 pp, medium confidence
47	Claude Opus 4.7 (Adaptive)	76.8%	estimated ± 2.6 pp, medium confidence
48	GPT-5.5	76.8%	estimated ± 2.6 pp, medium confidence
49	Gemini 3.5 Flash	76.8%	estimated ± 2.6 pp, medium confidence
50	Gemini 3.6 Flash	76.4%	estimated ± 2.6 pp, medium confidence
51	GPT-5.4	76.2%	estimated ± 2.6 pp, medium confidence
52	MiniMax M3	76.2%	estimated ± 2.6 pp, medium confidence
53	Hy3 Preview	76.0%	estimated ± 2.6 pp, medium confidence
54	Muse Spark 1.1	76.0%	estimated ± 2.6 pp, medium confidence
55	Apodex 1.1	75.9%	estimated ± 2.6 pp, medium confidence
56	Apodex 1.1 Mini	75.9%	estimated ± 2.6 pp, medium confidence
57	Quasar 438B	75.8%	estimated ± 2.6 pp, medium confidence
58	Ling 3.0 Flash VL	75.6%	estimated ± 2.6 pp, medium confidence
59	Nemotron 3 Ultra	75.6%	estimated ± 2.6 pp, medium confidence
60	Inkling-Small	75.3%	estimated ± 2.6 pp, high confidence
61	MiMo-V2.5-Pro	75.3%	estimated ± 2.6 pp, high confidence
62	GLM-5.1	75.3%	estimated ± 2.6 pp, high confidence
63	Solar Pro 4	75.2%	estimated ± 2.6 pp, high confidence
64	Qwen3.7 Max	75.0%	measured
65	Grok 4.3	75.0%	estimated ± 2.6 pp, high confidence
66	Inkling	74.9%	estimated ± 2.6 pp, high confidence
67	Hy3	74.8%	estimated ± 2.6 pp, high confidence
68	Kimi K2.6	74.5%	estimated ± 2.6 pp, high confidence
69	Kimi K2.7 Code	74.5%	estimated ± 2.6 pp, high confidence
70	GLM-4.7	74.3%	estimated ± 2.6 pp, high confidence
71	GPT-5.4 mini	74.3%	estimated ± 2.6 pp, high confidence
72	Step 3.7 Flash	74.3%	estimated ± 2.6 pp, high confidence
73	MiniMax M2.7	74.2%	estimated ± 2.6 pp, high confidence
74	Ornith-1.5-9B	74.2%	estimated ± 8.6 pp, low confidence
75	Muse Spark	74.1%	estimated ± 2.6 pp, high confidence
76	Qwen3.6 Plus	74.0%	estimated ± 2.6 pp, high confidence
77	Qwen3.6-27B	73.9%	estimated ± 2.6 pp, high confidence
78	Gemini 3.5 Flash-Lite	73.9%	estimated ± 2.6 pp, high confidence
79	A.X K2	73.6%	estimated ± 2.6 pp, high confidence
80	BTL-4	73.5%	measured
81	GPT-5.4 nano	73.5%	estimated ± 2.6 pp, high confidence
82	Ling 3.0 Flash FP8	73.4%	estimated ± 2.6 pp, high confidence
83	GPT-5 (high)	73.0%	estimated ± 2.6 pp, high confidence
84	Ling 3.0 Flash	73.0%	measured
85	Qwen3.7 Plus	72.9%	measured
86	Qwen3.6-35B-A3B	72.7%	estimated ± 2.6 pp, high confidence
87	Kimi K2.5	71.8%	estimated ± 2.6 pp, high confidence
88	Kimi K2.5 (Reasoning)	71.8%	estimated ± 2.6 pp, high confidence
89	Qwen 3.6 Max (preview)	71.5%	estimated ± 9.8 pp, low confidence
90	GPT-5.1	71.5%	estimated ± 2.6 pp, high confidence
91	Qwen3.5-122B-A10B	71.2%	estimated ± 2.6 pp, high confidence
92	Pokee-Isaac 28B	70.9%	measured
93	Gemini 3.1 Pro	70.7%	estimated ± 2.6 pp, high confidence
94	Muse Glimmer 30B	70.6%	estimated ± 2.6 pp, high confidence
95	Mistral Medium 3.5 128B	70.0%	estimated ± 2.6 pp, high confidence
96	Qwen3.5-27B	69.4%	estimated ± 9.8 pp, low confidence
97	Grok 4.1 Fast (Reasoning)	68.8%	estimated ± 9.8 pp, low confidence
98	Claude Opus 4.6 (Adaptive)	67.5%	estimated ± 9.8 pp, low confidence
99	GPT-5.2-Codex	67.5%	estimated ± 9.8 pp, low confidence
100	MiniCPM5-2B	66.6%	measured
101	Nemotron 3.5 Lightning 30B A3B NVFP4	65.5%	estimated ± 2.6 pp, high confidence
102	MiMo-V2-Flash	65.1%	estimated ± 2.6 pp, high confidence
103	Claude Opus 4.5 Thinking	64.9%	estimated ± 9.8 pp, low confidence
104	Qwen3.5-35B-A3B	64.6%	estimated ± 9.8 pp, low confidence
105	Gemma 4 31B	64.6%	estimated ± 2.6 pp, high confidence
106	Ornith-1.0-9B	64.6%	estimated ± 8.6 pp, low confidence
107	GPT-OSS 120B	64.0%	estimated ± 2.6 pp, high confidence
108	Gemini 3 Pro	62.6%	estimated ± 9.8 pp, low confidence
109	MiMo-V2.5	62.3%	estimated ± 8.6 pp, low confidence
110	GPT-5 (medium)	62.0%	estimated ± 9.8 pp, low confidence
111	GPT-5.3 Codex	61.6%	estimated ± 9.8 pp, low confidence
112	Granite 4.2 30B	61.4%	measured
113	Gemma 4 26B A4B	61.0%	estimated ± 2.6 pp, high confidence
114	LLaDA2.2-flash	60.8%	measured
115	Ling 3.0 Tiny	60.7%	estimated ± 2.6 pp, high confidence
116	GPT-5.2	60.5%	estimated ± 9.8 pp, low confidence
117	Qwen3.5 397B (Reasoning)	59.7%	estimated ± 9.8 pp, low confidence
118	GPT-5.1-Codex	59.0%	estimated ± 9.8 pp, low confidence
119	GPT-5.1-Codex-Max	59.0%	estimated ± 9.8 pp, low confidence
120	o3	57.2%	estimated ± 9.8 pp, low confidence
121	LFM2.5-2.6B	56.9%	measured
122	Ternary Bonsai 2 27B	56.8%	estimated ± 9.8 pp, low confidence
123	Command A+	56.2%	estimated ± 2.6 pp, high confidence
124	Agents-A1-4B	55.3%	estimated ± 9.8 pp, low confidence
125	Claude Opus 4.5	54.6%	estimated ± 8.6 pp, low confidence
126	GLM-4.6	54.5%	estimated ± 9.8 pp, low confidence
127	Celeris-1	53.8%	estimated ± 2.6 pp, high confidence
128	DeepSeek V3	53.8%	estimated ± 2.6 pp, high confidence
129	DeepSeek V3 0324	53.8%	estimated ± 2.6 pp, high confidence
130	Gemini 2.5 Pro	53.8%	estimated ± 2.6 pp, high confidence
131	Gemma 3 27B	53.8%	estimated ± 2.6 pp, high confidence
132	Gemma 4 12B	53.8%	estimated ± 2.6 pp, high confidence
133	Gemma 4 E2B	53.8%	estimated ± 2.6 pp, high confidence
134	Gemma 4 E4B	53.8%	estimated ± 2.6 pp, high confidence
135	GPT-4.1 mini	53.8%	estimated ± 2.6 pp, high confidence
136	GPT-4.1 nano	53.8%	estimated ± 2.6 pp, high confidence
137	GPT-4o	53.8%	estimated ± 2.6 pp, high confidence
138	GPT-4o mini	53.8%	estimated ± 2.6 pp, high confidence
139	GPT-OSS 20B	53.8%	estimated ± 2.6 pp, high confidence
140	K-Exaone	53.8%	estimated ± 2.6 pp, high confidence
141	Ling 2.6 Flash	53.8%	estimated ± 2.6 pp, high confidence
142	Llama 4 Maverick	53.8%	estimated ± 2.6 pp, high confidence
143	Llama 4 Scout	53.8%	estimated ± 2.6 pp, high confidence
144	Mercury 2.5	53.8%	estimated ± 2.6 pp, high confidence
145	Mistral Large 3	53.8%	estimated ± 2.6 pp, high confidence
146	Mistral Small 4	53.8%	estimated ± 2.6 pp, high confidence
147	Mistral Small 4 (Reasoning)	53.8%	estimated ± 2.6 pp, high confidence
148	Nemotron 3 Nano 30B	53.8%	estimated ± 2.6 pp, high confidence
149	Nemotron 3 Nano Omni 30B A3B	53.8%	estimated ± 2.6 pp, high confidence
150	Nemotron 3 Super 100B	53.8%	estimated ± 2.6 pp, high confidence
151	North Mini Code	53.8%	estimated ± 2.6 pp, high confidence
152	Solar Pro 3	53.8%	estimated ± 2.6 pp, high confidence
153	Trinity-Large-Preview	53.8%	estimated ± 2.6 pp, high confidence
154	Trinity-Large-Thinking	53.8%	estimated ± 2.6 pp, high confidence
155	Ultravox v0.6 Llama 3.3 70B	53.8%	estimated ± 2.6 pp, high confidence
156	Grok Code Fast 1	53.7%	estimated ± 9.8 pp, low confidence
157	Grok 4	53.2%	estimated ± 9.8 pp, low confidence
158	Qwen3 Max	52.9%	estimated ± 9.8 pp, low confidence
159	Claude Opus 4.7	52.7%	estimated ± 9.8 pp, low confidence
160	Granite 4.2 3B	52.4%	measured
161	Granite 4.2 8B	52.4%	measured
162	Claude 4.1 Opus Thinking	51.3%	estimated ± 9.8 pp, low confidence
163	LFM2.5-8B-A1B	49.7%	measured
164	MiMo-V2-Pro	49.6%	estimated ± 8.6 pp, low confidence
165	GLM-5	49.3%	estimated ± 8.6 pp, low confidence
166	Grok 4 Fast (Reasoning)	49.0%	estimated ± 9.8 pp, low confidence
167	Grok 4.1 Fast	48.3%	estimated ± 9.8 pp, low confidence
168	o1	48.0%	estimated ± 9.8 pp, low confidence
169	LLaDA2.2-mini	47.7%	measured
170	Kimi K2	47.6%	estimated ± 9.8 pp, low confidence
171	Qwen3.5 397B	46.7%	estimated ± 8.6 pp, low confidence
172	Claude 4 Sonnet	46.0%	estimated ± 9.8 pp, low confidence
173	Mellum2-12B-A2.5B-Thinking	45.6%	measured
174	GPT-4.1	45.5%	estimated ± 9.8 pp, low confidence
175	Sarvam 105B	45.5%	estimated ± 9.8 pp, low confidence
176	GLM-4.5-Air	45.5%	estimated ± 9.8 pp, low confidence
177	DeepSeek V3.1 (Reasoning)	45.1%	estimated ± 9.8 pp, low confidence
178	DeepSeek-R1	45.0%	estimated ± 9.8 pp, low confidence
179	DeepSeek V3.1	45.0%	estimated ± 9.8 pp, low confidence
180	Sarvam 30B	45.0%	estimated ± 9.8 pp, low confidence
181	Solar Pro 2	45.0%	estimated ± 9.8 pp, low confidence
182	Mistral Large 2	45.0%	estimated ± 9.8 pp, low confidence
183	o3-mini	44.9%	estimated ± 9.8 pp, low confidence
184	Mistral Medium 3	44.9%	estimated ± 9.8 pp, low confidence
185	Qwen3-Omni-30B-A3B-Thinking	44.9%	estimated ± 9.8 pp, low confidence
186	Claude 3 Haiku	44.9%	estimated ± 9.8 pp, low confidence
187	Exaone 4.0 1.2B	44.9%	estimated ± 9.8 pp, low confidence
188	Granite-4.0-H-1B	44.9%	estimated ± 9.8 pp, low confidence
189	Llama 3.1 405B	44.9%	estimated ± 9.8 pp, low confidence
190	Qwen3-Omni-30B-A3B-Instruct	44.9%	estimated ± 9.8 pp, low confidence
191	Gemini 2.5 Flash	44.9%	estimated ± 9.8 pp, low confidence
192	Granite-4.0-H-350M	44.9%	estimated ± 9.8 pp, low confidence
193	Nova Pro	44.9%	estimated ± 9.8 pp, low confidence
194	Granite-4.0-350M	44.9%	estimated ± 9.8 pp, low confidence
195	Nemotron Ultra 253B	44.9%	estimated ± 9.8 pp, low confidence
196	Exaone 4.0 32B	44.9%	estimated ± 9.8 pp, low confidence
197	LFM2.5-VL-1.6B-Extract	44.9%	estimated ± 9.8 pp, low confidence
198	Phi-4	44.9%	estimated ± 9.8 pp, low confidence
199	Mellum2-12B-A2.5B-Instruct	44.2%	measured
200	GLM-5-Turbo	43.9%	estimated ± 8.6 pp, low confidence
201	ZAYA1-8B	39.2%	measured
202	GLM-5V-Turbo	38.2%	estimated ± 8.6 pp, low confidence
203	LFM2.5-VL-3B	32.5%	measured
204	Gemini 3 Flash	25.2%	estimated ± 8.6 pp, low confidence
205	MiniCPM5-1B	25.2%	measured
206	LFM2.5-VL-450M	21.1%	measured
207	LFM2.5-230M	21.0%	measured
208	MiMo-V2-Omni	13.9%	estimated ± 8.6 pp, low confidence
209	DeepSeek V3.2	0.0%	estimated ± 8.6 pp, low confidence

BFCL v4 leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General