benchgap
Agentic · tools

BFCL v4 leaderboard

As of 2026-11-22, the highest measured score on BFCL v4 is 88.5% by BTL-3. 195 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#ModelScoreSource
1BTL-388.5%measured
2Claude Fable 5.177.4%estimated ± 2.7 pp, medium confidence
3Claude Opus 577.4%estimated ± 2.7 pp, medium confidence
4Inkling 2 Small77.3%estimated ± 4.7 pp, low confidence
5Grok 4.777.3%estimated ± 2.7 pp, medium confidence
6MiMo-V2.6-Pro77.2%estimated ± 2.7 pp, medium confidence
7ERNIE 5.5 Flash Preview77.2%estimated ± 4.7 pp, low confidence
8Muse Spark 1.377.2%estimated ± 2.7 pp, medium confidence
9Qwen3.8 Max Preview77.2%estimated ± 2.7 pp, medium confidence
10GLM-5.377.2%estimated ± 2.7 pp, medium confidence
11Qwen3.8-Flash-Next77.1%estimated ± 2.7 pp, medium confidence
12Gemini 4 Argon77.1%estimated ± 2.7 pp, medium confidence
13Claude Haiku 5.577.1%estimated ± 2.7 pp, medium confidence
14Grok 4.677.1%estimated ± 2.7 pp, medium confidence
15Ling 3.1 Flash77.1%estimated ± 2.7 pp, medium confidence
16Claude Fable 577.0%estimated ± 2.7 pp, medium confidence
17GPT-5.6 Sol77.0%estimated ± 2.7 pp, medium confidence
18MiMo-V2.6-Flash77.0%estimated ± 2.7 pp, medium confidence
19Hy4 preview77.0%estimated ± 4.7 pp, low confidence
20Atria Dawn Preview77.0%measured
21DeepSeek V4.1 Flash 101777.0%estimated ± 2.7 pp, medium confidence
22GPT-6.1 Sol77.0%estimated ± 2.7 pp, medium confidence
23DeepSeek V4 Pro77.0%estimated ± 2.7 pp, medium confidence
24Step 5 Preview77.0%estimated ± 2.7 pp, medium confidence
25GPT-6 Astra76.9%estimated ± 2.7 pp, medium confidence
26LongCat-2.576.8%estimated ± 4.7 pp, low confidence
27Kimi K376.8%estimated ± 2.7 pp, medium confidence
28Inkling 2 Mini76.7%estimated ± 4.7 pp, low confidence
29GPT-6 Sol76.7%estimated ± 2.7 pp, medium confidence
30Muse Spark 1.276.5%estimated ± 2.7 pp, medium confidence
31Claude Sonnet 576.5%estimated ± 2.7 pp, medium confidence
32GPT-5.6 Luna76.5%estimated ± 2.7 pp, medium confidence
33MiniMax M3.1 Flash76.5%estimated ± 2.7 pp, medium confidence
34Claude Opus 4.876.5%estimated ± 2.7 pp, medium confidence
35GPT-5.6 Terra76.5%estimated ± 2.7 pp, medium confidence
36Beam 1.0 501B-A23B FP876.4%estimated ± 4.7 pp, low confidence
37Gemini 3.8 Flash76.4%estimated ± 2.7 pp, medium confidence
38DeepSeek V4 Flash 073176.4%estimated ± 2.7 pp, medium confidence
39GPT-6 Luna76.4%estimated ± 2.7 pp, medium confidence
40Mistral Large 476.3%estimated ± 2.7 pp, medium confidence
41Qwen3.8-27B76.3%estimated ± 2.7 pp, medium confidence
42Muse Spark Open76.2%estimated ± 2.7 pp, medium confidence
43Gemini 3.7 Flash76.2%estimated ± 2.7 pp, medium confidence
44Grok 4.576.2%estimated ± 2.7 pp, medium confidence
45Beam 1.0 501B-A23B NVFP476.1%estimated ± 4.7 pp, low confidence
46GLM-5.276.1%estimated ± 2.7 pp, medium confidence
47Claude Opus 4.776.0%estimated ± 2.7 pp, medium confidence
48GPT-5.576.0%estimated ± 2.7 pp, medium confidence
49Ornith-1.5-397B76.0%estimated ± 4.7 pp, low confidence
50Gemini 3.5 Flash76.0%estimated ± 2.7 pp, medium confidence
51Gemini 3.6 Flash75.7%estimated ± 2.7 pp, medium confidence
52GPT-5.475.4%estimated ± 2.7 pp, medium confidence
53MiniMax M375.4%estimated ± 2.7 pp, medium confidence
54Hy3 Preview75.2%estimated ± 2.7 pp, medium confidence
55Muse Spark 1.175.2%estimated ± 2.7 pp, medium confidence
56Apodex 1.175.1%estimated ± 2.7 pp, medium confidence
57Apodex 1.1 Mini75.1%estimated ± 2.7 pp, medium confidence
58Quasar 438B75.0%estimated ± 2.7 pp, medium confidence
59Qwen3.7 Max75.0%measured
60Beam 1.0 501B-A23B75.0%estimated ± 2.7 pp, medium confidence
61Nemotron 3.5 Super VL 120B A12B74.9%estimated ± 2.7 pp, medium confidence
62Nemotron 3.5 Super VL 120B A12B NVFP474.9%estimated ± 2.7 pp, medium confidence
63Nemotron 3 Ultra74.9%estimated ± 2.7 pp, medium confidence
64Ling 3.0 Flash VL74.9%estimated ± 2.7 pp, medium confidence
65Ling 3.0 Flash Sante74.6%estimated ± 4.7 pp, low confidence
66Inkling-Small74.6%estimated ± 2.7 pp, high confidence
67MiMo-V2.5-Pro74.5%estimated ± 2.7 pp, high confidence
68GLM-5.174.5%estimated ± 2.7 pp, high confidence
69Solar Pro 474.4%estimated ± 2.7 pp, high confidence
70Grok 4.374.2%estimated ± 2.7 pp, high confidence
71Inkling74.1%estimated ± 2.7 pp, high confidence
72Hy374.1%estimated ± 2.7 pp, high confidence
73Kimi K2.673.8%estimated ± 2.7 pp, high confidence
74Kimi K2.7 Code73.8%estimated ± 2.7 pp, high confidence
75Step 3.7 Flash73.6%estimated ± 2.7 pp, high confidence
76GLM-4.773.6%estimated ± 2.7 pp, high confidence
77GPT-5.4 mini73.5%estimated ± 2.7 pp, high confidence
78MiniMax M2.773.5%estimated ± 2.7 pp, high confidence
79BTL-473.5%measured
80Ripple-65B Preview73.4%estimated ± 4.7 pp, low confidence
81Muse Spark73.4%estimated ± 2.7 pp, high confidence
82Qwen3.6 Plus73.3%estimated ± 2.7 pp, high confidence
83Qwen3.6-27B73.3%estimated ± 2.7 pp, high confidence
84Gemini 3.5 Flash-Lite73.2%estimated ± 2.7 pp, high confidence
85Nemotron 3.5 Bolt 74B-A9B73.2%estimated ± 2.7 pp, high confidence
86Ling 3.0 Flash73.0%measured
87Ornith-1.5-35B-A3B72.9%estimated ± 4.7 pp, low confidence
88A.X K272.9%estimated ± 2.7 pp, high confidence
89Qwen3.7 Plus72.9%measured
90GPT-5.4 nano72.8%estimated ± 2.7 pp, high confidence
91GPT-572.4%estimated ± 2.7 pp, high confidence
92Qwen3.6-35B-A3B72.0%estimated ± 2.7 pp, high confidence
93Spark-X2.6-4B71.8%measured
94LFM3-1.7B71.4%measured
95Pokee-Isaac 28B70.9%measured
96GPT-5.170.8%estimated ± 2.7 pp, high confidence
97Qwen3.5-122B-A10B70.6%estimated ± 2.7 pp, high confidence
98GLM-5-Turbo70.6%estimated ± 8.5 pp, low confidence
99GLM-5V-Turbo70.6%estimated ± 8.5 pp, low confidence
100K-EXAONE 2.070.5%estimated ± 2.7 pp, high confidence
101Gemini 3.1 Pro70.1%estimated ± 2.7 pp, high confidence
102Muse Glimmer 30B69.9%estimated ± 2.7 pp, high confidence
103Solar Open 269.6%estimated ± 4.7 pp, low confidence
104Qwen 3.6 Max (preview)69.4%estimated ± 8.5 pp, low confidence
105Mistral Medium 3.5 128B69.4%estimated ± 2.7 pp, high confidence
106MiMo-V2-Pro68.9%estimated ± 8.5 pp, low confidence
107Ornith-1.5-9B68.9%estimated ± 4.7 pp, low confidence
108Qwen3.5-27B68.4%estimated ± 8.5 pp, low confidence
109Kimi K2.568.3%measured
110Grok 4.1 Fast (Reasoning)68.1%estimated ± 8.5 pp, low confidence
111Qwen3.5 397B67.8%estimated ± 4.7 pp, low confidence
112LongCat-Flash-Lite-Sparse67.8%estimated ± 4.7 pp, low confidence
113Claude Opus 4.567.5%estimated ± 4.7 pp, low confidence
114Claude Opus 4.6 (Adaptive)67.5%estimated ± 8.5 pp, low confidence
115GPT-5.2-Codex67.5%estimated ± 8.5 pp, low confidence
116Qwen3.5-4B67.5%estimated ± 8.5 pp, low confidence
117GLM-567.2%estimated ± 4.7 pp, low confidence
118MiMo-V2-Omni67.1%estimated ± 8.5 pp, low confidence
119Nemotron 3.5 Bolt 74B-A9B BF1666.9%estimated ± 2.9 pp, medium confidence
120Sage-A1.566.7%estimated ± 2.9 pp, medium confidence
121MiniCPM5-2B66.6%measured
122Claude Opus 4.5 Thinking66.3%estimated ± 8.5 pp, low confidence
123Qwen3.5-35B-A3B66.1%estimated ± 8.5 pp, low confidence
124Nemotron 3.5 Lightning 30B A3B NVFP465.1%estimated ± 2.7 pp, high confidence
125Gemini 3 Pro65.1%estimated ± 8.5 pp, low confidence
126LFM2.5-24B-A2B65.0%measured
127MiMo-V2-Flash64.6%estimated ± 2.7 pp, high confidence
128GPT-5.3 Codex64.5%estimated ± 8.5 pp, low confidence
129Gemma 4 31B64.1%estimated ± 2.7 pp, high confidence
130Claude Opus 4.663.9%estimated ± 8.5 pp, low confidence
131GPT-5.263.9%estimated ± 8.5 pp, low confidence
132GPT-OSS 120B63.5%estimated ± 2.7 pp, high confidence
133GPT-5.1-Codex63.0%estimated ± 8.5 pp, low confidence
134GPT-5.1-Codex-Max63.0%estimated ± 8.5 pp, low confidence
135o361.8%estimated ± 8.5 pp, low confidence
136Ternary Bonsai 2 27B61.5%estimated ± 8.5 pp, low confidence
137Granite 4.2 30B61.4%measured
138Claude Sonnet 4.661.1%estimated ± 8.5 pp, low confidence
139DeepSeek V3.260.8%estimated ± 8.5 pp, low confidence
140LLaDA2.2-flash60.8%measured
141Gemma 4 26B A4B60.5%estimated ± 2.7 pp, high confidence
142Agents-A1-4B60.4%estimated ± 8.5 pp, low confidence
143Ling 3.0 Tiny60.2%estimated ± 2.7 pp, high confidence
144GLM-4.659.7%estimated ± 8.5 pp, low confidence
145Grok Code Fast 159.1%estimated ± 8.5 pp, low confidence
146Grok 458.6%estimated ± 8.5 pp, low confidence
147Qwen3 Max58.3%estimated ± 8.5 pp, low confidence
148Claude 4 Opus Thinking57.8%estimated ± 8.5 pp, low confidence
149LFM2.5-2.6B56.9%measured
150Command A+55.7%estimated ± 2.7 pp, high confidence
151Celeris-153.9%estimated ± 2.7 pp, high confidence
152Claude 3.5 Haiku53.9%estimated ± 2.7 pp, high confidence
153DeepSeek V353.9%estimated ± 2.7 pp, high confidence
154DeepSeek V3 032453.9%estimated ± 2.7 pp, high confidence
155Gemini 2.5 Pro53.9%estimated ± 2.7 pp, high confidence
156Gemma 3 27B53.9%estimated ± 2.7 pp, high confidence
157Gemma 4 12B53.9%estimated ± 2.7 pp, high confidence
158Gemma 4 E2B53.9%estimated ± 2.7 pp, high confidence
159Gemma 4 E4B53.9%estimated ± 2.7 pp, high confidence
160GPT-4.1 mini53.9%estimated ± 2.7 pp, high confidence
161GPT-4.1 nano53.9%estimated ± 2.7 pp, high confidence
162GPT-4o53.9%estimated ± 2.7 pp, high confidence
163GPT-4o mini53.9%estimated ± 2.7 pp, high confidence
164GPT-OSS 20B53.9%estimated ± 2.7 pp, high confidence
165K-Exaone53.9%estimated ± 2.7 pp, high confidence
166Ling 2.6 Flash53.9%estimated ± 2.7 pp, high confidence
167Llama 3.3 70B53.9%estimated ± 2.7 pp, high confidence
168Llama 4 Maverick53.9%estimated ± 2.7 pp, high confidence
169Llama 4 Scout53.9%estimated ± 2.7 pp, high confidence
170Mercury 2.553.9%estimated ± 2.7 pp, high confidence
171Mistral Large 353.9%estimated ± 2.7 pp, high confidence
172Mistral Small 3.153.9%estimated ± 2.7 pp, high confidence
173Mistral Small 453.9%estimated ± 2.7 pp, high confidence
174Nemotron 3 Nano 30B53.9%estimated ± 2.7 pp, high confidence
175Nemotron 3 Nano Omni 30B A3B53.9%estimated ± 2.7 pp, high confidence
176North Mini Code53.9%estimated ± 2.7 pp, high confidence
177Solar Pro 353.9%estimated ± 2.7 pp, high confidence
178Trinity-Large-Preview53.9%estimated ± 2.7 pp, high confidence
179Trinity-Large-Thinking53.9%estimated ± 2.7 pp, high confidence
180Ultravox v0.6 Llama 3.3 70B53.9%estimated ± 2.7 pp, high confidence
181Grok 4 Fast (Reasoning)53.4%estimated ± 8.5 pp, low confidence
182Granite 4.2 3B52.4%measured
183Granite 4.2 8B52.4%measured
184Grok 4.1 Fast52.1%estimated ± 8.5 pp, low confidence
185o151.5%estimated ± 8.5 pp, low confidence
186Kimi K250.6%estimated ± 8.5 pp, low confidence
187LFM2.5-8B-A1B49.7%measured
188LLaDA2.2-mini47.7%measured
189o4-mini47.1%estimated ± 8.5 pp, low confidence
190Nemotron 3.5 Bolt E4B BF1646.7%estimated ± 8.5 pp, low confidence
191Claude 3.7 Sonnet Thinking46.5%estimated ± 8.5 pp, low confidence
192Mellum2-12B-A2.5B-Thinking45.6%measured
193Nemotron 3.5 Bolt E4B44.9%estimated ± 8.5 pp, low confidence
194Claude 4 Sonnet44.9%estimated ± 8.5 pp, low confidence
195Mellum2-12B-A2.5B-Instruct44.2%measured
196Claude 3.7 Sonnet43.4%estimated ± 8.5 pp, low confidence
197GPT-4.141.4%estimated ± 8.5 pp, low confidence
198Sarvam 105B41.2%estimated ± 8.5 pp, low confidence
199GLM-4.5-Air40.9%estimated ± 8.5 pp, low confidence
200ZAYA1-8B39.2%measured
201Gemini 3 Flash38.7%estimated ± 8.5 pp, low confidence
202DeepSeek-R133.6%estimated ± 8.5 pp, low confidence
203LFM2.5-VL-3B32.5%measured
204DeepSeek V3.132.3%estimated ± 8.5 pp, low confidence
205Sarvam 30B32.1%estimated ± 8.5 pp, low confidence
206Solar Pro 230.0%estimated ± 8.5 pp, low confidence
207Mistral Large 229.1%estimated ± 8.5 pp, low confidence
208Gemini 2.0 Flash28.1%estimated ± 8.5 pp, low confidence
209o3-mini27.4%estimated ± 8.5 pp, low confidence
210Devstral Small27.2%estimated ± 8.5 pp, low confidence
211Magistral Small25.7%estimated ± 8.5 pp, low confidence
212MiniCPM5-1B25.2%measured
213Mistral Medium 323.7%estimated ± 8.5 pp, low confidence
214Magistral Medium22.7%estimated ± 8.5 pp, low confidence
215Llama 3.1 70B21.4%estimated ± 8.5 pp, low confidence
216Qwen3-Omni-30B-A3B-Thinking21.1%estimated ± 8.5 pp, low confidence
217LFM2.5-VL-450M21.1%measured
218LFM2.5-230M21.0%measured
219Claude 3 Haiku20.9%estimated ± 8.5 pp, low confidence
220Exaone 4.0 1.2B20.4%estimated ± 8.5 pp, low confidence
221Granite-4.0-H-1B19.6%estimated ± 8.5 pp, low confidence

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General