Coding

Vibe Code Bench leaderboard

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

As of 2026-10-07, the highest measured score on Vibe Code Bench is 91.9% by Gemini 4 Argon. 135 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Claude Fable 5.1	100.0%	estimated ± 10.3 pp, low confidence
2	Gemini 4 Argon	91.9%	measured
3	Claude Opus 5.5	81.7%	estimated ± 5.6 pp, low confidence
4	Claude Fable 5	81.0%	estimated ± 5.6 pp, low confidence
5	Claude Opus 5	80.9%	estimated ± 5.6 pp, low confidence
6	GPT-6 Astra	80.8%	estimated ± 5.6 pp, low confidence
7	Mistral Large 4	78.4%	measured
8	SWE-2	78.2%	estimated ± 5.6 pp, low confidence
9	GPT-5.6 Sol	77.7%	estimated ± 10.3 pp, low confidence
10	Claude Opus 4.8	75.2%	estimated ± 5.6 pp, low confidence
11	Claude Haiku 5.5	75.2%	estimated ± 5.6 pp, low confidence
12	Grok 4.6	75.0%	estimated ± 10.3 pp, low confidence
13	Claude Sonnet 5.5	75.0%	estimated ± 5.6 pp, low confidence
14	GPT-5.6 Terra	74.5%	estimated ± 10.3 pp, low confidence
15	Gemini 3.8 Flash	72.9%	estimated ± 10.3 pp, low confidence
16	Kimi K3	72.7%	estimated ± 10.3 pp, low confidence
17	Gemini 3.7 Flash	72.6%	estimated ± 5.6 pp, low confidence
18	Claude Sonnet 5	71.8%	estimated ± 5.6 pp, low confidence
19	SWE-1.7	71.4%	estimated ± 5.6 pp, low confidence
20	Claude Opus 4.7	71.0%	measured
21	Muse Spark 1.3	70.9%	estimated ± 10.3 pp, low confidence
22	GPT-5.5	69.8%	measured
23	GPT-5.4	67.4%	measured
24	GLM-5.3	67.0%	estimated ± 10.3 pp, low confidence
25	Composer 2	66.5%	estimated ± 6.5 pp, low confidence
26	Composer 2 Fast	66.5%	estimated ± 6.5 pp, low confidence
27	Claude Mythos 5	62.5%	estimated ± 8.1 pp, low confidence
28	Ember-1	62.5%	estimated ± 8.1 pp, low confidence
29	Claude Opus 4.7 (Adaptive)	62.4%	estimated ± 8.1 pp, low confidence
30	Ornith-1.5-397B	62.3%	estimated ± 8.1 pp, low confidence
31	GPT-5.3 Codex	61.8%	measured
32	Qwen3.8-Flash-Next	61.2%	estimated ± 10.3 pp, low confidence
33	Ornith-1.0-397B	59.8%	estimated ± 8.1 pp, medium confidence
34	Grok 4.5	59.4%	estimated ± 10.3 pp, low confidence
35	Muse Spark 1.2	58.7%	estimated ± 10.3 pp, low confidence
36	Claude Opus 4.6	57.6%	measured
37	Qwen3.8 Max Preview	57.5%	estimated ± 10.3 pp, low confidence
38	GPT-5.6 Luna	56.5%	estimated ± 10.3 pp, low confidence
39	Muse Spark 1.1	56.1%	estimated ± 10.3 pp, low confidence
40	Claude Opus 4.5	54.1%	estimated ± 8.1 pp, medium confidence
41	Beam	54.1%	estimated ± 8.1 pp, medium confidence
42	GPT-5.2	53.5%	measured
43	Claude Opus 4.6 (Adaptive)	53.5%	measured
44	Claude Sonnet 4.6	51.5%	measured
45	MiniMax M3	51.4%	estimated ± 8.1 pp, medium confidence
46	Gemini 3.6 Flash	50.7%	estimated ± 10.3 pp, low confidence
47	Qwen3.7 Max	50.6%	estimated ± 8.1 pp, medium confidence
48	DeepSeek V4 Pro 0813	49.9%	measured
49	GLM-5.2	49.5%	estimated ± 10.3 pp, low confidence
50	Inkling-Small	48.9%	estimated ± 8.1 pp, medium confidence
51	Gemini 3.5 Flash	48.7%	measured
52	Qwen3.8-27B	48.0%	estimated ± 10.3 pp, low confidence
53	GPT-5.4 mini	48.0%	measured
54	GPT-5.2-Codex	37.9%	measured
55	Kimi K2.6	37.9%	measured
56	DeepSeek V4 Flash 0731	35.7%	estimated ± 8.1 pp, medium confidence
57	Ornith-1.5-35B-A3B	35.7%	estimated ± 8.1 pp, medium confidence
58	Quasar 438B	35.4%	estimated ± 10.3 pp, low confidence
59	Apodex 1.1 Mini	34.7%	estimated ± 10.3 pp, low confidence
60	Kimi K2.7 Code	34.7%	estimated ± 10.3 pp, low confidence
61	MiMo-V2.5-Pro	33.9%	estimated ± 10.3 pp, low confidence
62	Gemini 3.1 Pro	32.0%	measured
63	Hy3	32.0%	estimated ± 10.3 pp, low confidence
64	GLM-5.1	31.5%	measured
65	BTL-4	28.1%	estimated ± 8.1 pp, medium confidence
66	dots3-note Preview	28.1%	estimated ± 8.1 pp, medium confidence
67	Kimi K2.5	27.9%	estimated ± 6.5 pp, low confidence
68	Gemma 4 31B	27.7%	estimated ± 6.5 pp, low confidence
69	GLM-5	27.7%	estimated ± 6.5 pp, low confidence
70	Grok 4	27.7%	estimated ± 6.5 pp, low confidence
71	DeepSeek V3.2	27.7%	estimated ± 6.5 pp, low confidence
72	GPT-OSS 120B	27.7%	estimated ± 6.5 pp, low confidence
73	GPT-OSS 20B	27.7%	estimated ± 6.5 pp, low confidence
74	MiniMax M2.7	27.0%	measured
75	GPT-5.4 nano	26.1%	measured
76	Qwen3.6 Plus	25.6%	measured
77	GPT-5.1	24.6%	measured
78	GLM-5 (Reasoning)	23.4%	measured
79	MiMo-V2-Pro	23.1%	estimated ± 8.1 pp, medium confidence
80	Ling 3.0 Flash	22.9%	estimated ± 10.3 pp, low confidence
81	Ling 3.0 Flash FP8	22.9%	estimated ± 10.3 pp, low confidence
82	Claude Sonnet 4.5 Thinking	22.6%	measured
83	GPT-5.1-Codex-Max	22.2%	measured
84	Gemini 3.5 Flash-Lite	21.7%	estimated ± 10.3 pp, low confidence
85	Claude Opus 4.5 Thinking	20.6%	measured
86	Gemini 3 Flash	20.2%	measured
87	GPT-5 (high)	20.1%	measured
88	Apodex 1.1	19.7%	estimated ± 8.1 pp, medium confidence
89	Qwen3.7 Plus	19.7%	estimated ± 8.1 pp, medium confidence
90	Muse Spark	19.7%	measured
91	Inkling	18.6%	estimated ± 8.1 pp, medium confidence
92	Mistral Medium 3.5 128B	18.6%	estimated ± 8.1 pp, medium confidence
93	Kimi K2.5 (Reasoning)	17.5%	measured
94	Grok 4.3	16.3%	estimated ± 10.3 pp, low confidence
95	Qwen3.5 Plus	15.7%	measured
96	MiniMax M2.5	14.9%	measured
97	Claude Sonnet 4.5	14.7%	estimated ± 8.1 pp, medium confidence
98	Qwen3.6-27B	14.7%	estimated ± 8.1 pp, medium confidence
99	o1	14.7%	estimated ± 10.3 pp, low confidence
100	Step 3.7 Flash	14.6%	estimated ± 10.3 pp, low confidence
101	Gemma 4 26B A4B	14.4%	estimated ± 10.3 pp, low confidence
102	Gemini 3 Pro	14.3%	measured
103	GPT-5 mini	14.2%	measured
104	Nemotron 3 Super 100B	13.5%	estimated ± 10.3 pp, low confidence
105	GPT-5.1-Codex	13.1%	measured
106	o1-preview	11.5%	estimated ± 10.3 pp, low confidence
107	Claude Haiku 4.5 Thinking	11.4%	measured
108	K-Exaone	10.5%	estimated ± 10.3 pp, low confidence
109	Gemma 4 12B	10.0%	estimated ± 10.3 pp, low confidence
110	Command A+	8.6%	estimated ± 10.3 pp, low confidence
111	Mistral Small 4	8.1%	estimated ± 10.3 pp, low confidence
112	Mistral Small 4 (Reasoning)	8.1%	estimated ± 10.3 pp, low confidence
113	Trinity-Large-Preview	7.7%	estimated ± 10.3 pp, low confidence
114	Trinity-Large-Thinking	7.7%	estimated ± 10.3 pp, low confidence
115	Qwen3.5 397B	7.7%	estimated ± 8.1 pp, medium confidence
116	Ling 2.6 Flash	7.5%	estimated ± 10.3 pp, low confidence
117	Gemini 1.5 Pro	6.9%	estimated ± 10.3 pp, low confidence
118	Muse Glimmer 30B	6.7%	estimated ± 8.1 pp, medium confidence
119	GPT-4 Turbo	6.1%	estimated ± 10.3 pp, low confidence
120	Mistral Large 3	5.6%	estimated ± 10.3 pp, low confidence
121	Claude 3 Opus	5.4%	estimated ± 10.3 pp, low confidence
122	Ornith-1.0-35B	5.1%	estimated ± 8.1 pp, medium confidence
123	DeepSeek V3.2 (Thinking)	5.1%	measured
124	Llama 4 Maverick	4.3%	estimated ± 10.3 pp, low confidence
125	Grok 4.20	4.1%	measured
126	Celeris-1	3.7%	estimated ± 10.3 pp, low confidence
127	Nemotron 3 Nano 30B	3.7%	estimated ± 10.3 pp, low confidence
128	Nemotron 3 Nano Omni 30B A3B	3.5%	estimated ± 10.3 pp, low confidence
129	Qwen3 Max	3.5%	measured
130	GLM-4.6	3.1%	measured
131	MiMo-V2-Omni	3.0%	estimated ± 8.1 pp, medium confidence
132	Ultravox v0.6 Llama 3.3 70B	3.0%	estimated ± 10.3 pp, low confidence
133	GPT-4o mini	2.9%	estimated ± 10.3 pp, low confidence
134	GPT-4.1 nano	2.8%	estimated ± 10.3 pp, low confidence
135	Laguna M.1	2.7%	estimated ± 8.1 pp, medium confidence
136	Claude 4.1 Opus	2.5%	estimated ± 8.1 pp, medium confidence
137	Gemma 3 27B	2.5%	estimated ± 10.3 pp, low confidence
138	Hy3 Preview	2.4%	estimated ± 8.1 pp, medium confidence
139	Gemma 4 E4B	2.3%	estimated ± 10.3 pp, low confidence
140	Llama 4 Scout	2.0%	estimated ± 10.3 pp, low confidence
141	LFM2.5-2.6B	1.9%	estimated ± 10.3 pp, low confidence
142	Gemma 4 E2B	1.7%	estimated ± 10.3 pp, low confidence
143	GLM-4.7	1.7%	estimated ± 8.1 pp, medium confidence
144	MAI-Thinking-1	1.5%	estimated ± 8.1 pp, medium confidence
145	MiMo-V2-Flash	1.4%	estimated ± 8.1 pp, medium confidence
146	Qwen3.6-35B-A3B	1.4%	estimated ± 8.1 pp, medium confidence
147	Claude Haiku 4.5	1.4%	estimated ± 8.1 pp, medium confidence
148	Grok 4.1 Fast (Reasoning)	1.2%	measured
149	Claude 4 Sonnet	1.1%	estimated ± 8.1 pp, medium confidence
150	MAI-Code-1.1-Flash	1.0%	estimated ± 8.1 pp, medium confidence
151	Qwen3.5-27B	1.0%	estimated ± 8.1 pp, medium confidence
152	Qwen3.5-122B-A10B	0.9%	estimated ± 8.1 pp, medium confidence
153	Nemotron 3 Ultra	0.9%	estimated ± 8.1 pp, medium confidence
154	Laguna XS 2.1	0.7%	estimated ± 8.1 pp, medium confidence
155	Grok Code Fast 1	0.7%	estimated ± 8.1 pp, medium confidence
156	Ornith-1.5-9B	0.7%	estimated ± 8.1 pp, medium confidence
157	Solar Pro 4	0.7%	estimated ± 8.1 pp, medium confidence
158	Solar Open 2	0.7%	estimated ± 8.1 pp, medium confidence
159	Laguna XS.2	0.7%	estimated ± 8.1 pp, medium confidence
160	Ornith-1.0-9B	0.7%	estimated ± 8.1 pp, medium confidence
161	Qwen3.5-35B-A3B	0.7%	estimated ± 8.1 pp, medium confidence
162	K-EXAONE 2.0	0.7%	estimated ± 8.1 pp, medium confidence
163	LongCat-Flash-Lite-Sparse	0.7%	estimated ± 8.1 pp, medium confidence
164	Claude 3.5 Sonnet	0.6%	estimated ± 8.1 pp, low confidence
165	DeepSeek V3	0.6%	estimated ± 8.1 pp, low confidence
166	GPT-4.1	0.6%	estimated ± 8.1 pp, low confidence
167	GPT-4.1 mini	0.6%	estimated ± 8.1 pp, low confidence
168	Granite 4.2 30B	0.6%	estimated ± 8.1 pp, low confidence
169	Granite 4.2 8B	0.6%	estimated ± 8.1 pp, low confidence
170	LLaDA2.2-flash	0.6%	estimated ± 8.1 pp, low confidence
171	MiniCPM5-2B	0.6%	estimated ± 8.1 pp, low confidence
172	Nemotron 3.5 Lightning 30B A3B NVFP4	0.6%	estimated ± 8.1 pp, low confidence
173	o3-mini	0.6%	estimated ± 8.1 pp, low confidence
174	Ternary Bonsai 2 27B	0.6%	estimated ± 8.1 pp, low confidence
175	ZAYA1-74B-Preview	0.6%	estimated ± 8.1 pp, low confidence
176	Gemini 2.5 Pro	0.4%	measured
177	Gemini 3.1 Flash-Lite	0.0%	measured
178	Grok 4 Fast (Reasoning)	0.0%	measured

Vibe Code Bench leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General