LiveCodeBench

LiveCodeBench: competitive programming problems; pass rate.

232models

–Open Index weight

–last updated

Artificial Analysispublisher

#	Model	Effort	Score	Cost / task	Harness
1	Gemini 3 Pro	high	91.7%	–	–
2	Gemini 3 Flash	–	90.8%	–	–
3	DeepSeek V3.2	–	89.6%	–	–
4	GPT-5.2	medium	89.4%	–	–
5	GLM 4.7	–	89.4%	–	–
6	gpt-oss-120b	high	87.8%	–	–
7	Opus 4.5	–	87.1%	–	–
8	GPT-5.1	high	86.8%	–	–
9	MiMo V2 Flash	–	86.8%	–	–
10	o4 mini	high	85.9%	–	–
11	Kimi K2	–	85.3%	–	–
12	GPT-5.1 Codex	high	84.9%	–	–
13	GPT-5	high	84.6%	–	–
14	GPT-5 Codex	high	84.0%	–	–
15	GPT-5 Mini	high	83.8%	–	–
16	GPT-5.1 Codex Mini	high	83.6%	–	–
17	Grok 4 Fast	–	83.2%	–	–
18	MiniMax M2	–	82.6%	–	–
19	Grok 4.1 Fast	–	82.2%	–	–
20	Grok 4	–	81.9%	–	–
21	ERNIE 5.0 Thinking Preview	–	81.2%	–	–
22	MiniMax M2.1	–	81.0%	–	–
23	o3	–	80.8%	–	–
24	Apriel-v1.6-15B-Thinker	–	80.7%	–	–
25	Gemini 2.5 Pro	–	80.1%	–	–
26	DeepSeek V3.1	–	79.8%	–	–
27	GPT-5 Nano	high	78.9%	–	–
28	Qwen 3 235B-A22B	–	78.8%	–	–
29	Qwen 3	–	78.4%	–	–
30	INTELLECT-3 (Based on GLM-4.5-Air)	–	77.7%	–	–
31	gpt-oss-20b	high	77.7%	–	–
32	DeepSeek R1	–	77.0%	–	–
33	K-EXAONE (Reasoning)	–	76.8%	–	–
34	Qwen 3 Max	none	76.7%	–	–
35	Doubao Seed Code	–	76.6%	–	–
36	Seed-OSS-36B-Instruct	–	76.5%	–	–
37	Magistral Medium 1.2	medium	75.0%	–	–
38	KAT-Coder-Pro V1	none	74.7%	–	–
39	EXAONE 4.0 32B (Reasoning)	–	74.7%	–	–
40	NVIDIA Nemotron 3 Nano 30B A3B (Reasoning)	–	74.1%	–	–
41	GLM 4.5	–	73.8%	–	–
42	Llama Nemotron Super 49B v1.5 (Reasoning)	–	73.7%	–	–
43	o3 mini	high	73.4%	–	–
44	Nova 2.0 Pro Preview (Medium)	medium	73.0%	–	–
45	Apriel-v1.5-15B-Thinker	–	72.8%	–	–
46	Falcon-H1R-7B	–	72.4%	–	–
47	NVIDIA Nemotron Nano 9B V2 (Reasoning)	–	72.4%	–	–
48	Magistral Small 1.2	–	72.3%	–	–
49	Sonnet 4.5	–	71.4%	–	–
50	Gemini 2.5 Flash	–	71.3%	–	–
51	Nova 2.0 Lite (High)	high	71.1%	–	–
52	MiniMax M1	–	71.1%	–	–
53	Qwen 3 30B-A3B	–	70.7%	–	–
54	NVIDIA Nemotron Nano 9B V2 (Non-reasoning)	none	70.1%	–	–
55	Grok 3 Mini	high	69.6%	–	–
56	GLM 4.6	–	69.5%	–	–
57	Olmo 3.1 32B Think	–	69.5%	–	–
58	K2-V2 (High)	high	69.4%	–	–
59	NVIDIA Nemotron Nano 12B v2 VL (Reasoning)	–	69.4%	–	–
60	Gemini 2.5 Flash Lite	–	68.8%	–	–
61	Cogito v2.1 (Reasoning)	–	68.8%	–	–
62	Llama 3.1 405b	–	68.6%	–	–
63	GLM 4.5 Air	–	68.4%	–	–
64	o1	–	67.9%	–	–
65	Ling-1T	none	67.7%	–	–
66	Olmo 3 32B Think	–	67.2%	–	–
67	Nova 2.0 Lite (Medium)	medium	66.3%	–	–
68	Nova 2.0 Omni (Medium)	medium	66.0%	–	–
69	Grok Code Fast 1	–	65.7%	–	–
70	Mi:dm K 2.5 Pro	–	65.6%	–	–
71	Sonnet 4	–	65.5%	–	–
72	Opus 4.1	–	65.4%	–	–
73	Llama 3.1 70b	–	65.3%	–	–
74	Motif-2-12.7B-Reasoning	–	65.1%	–	–
75	Ring-1T	–	64.3%	–	–
76	Qwen 3 4B	–	64.1%	–	–
77	Llama 3.1	max	64.1%	–	–
78	Nova 2.0 Pro Preview (Low)	low	63.8%	–	–
79	Opus 4	–	63.6%	–	–
80	QwQ 32B	–	63.1%	–	–
81	HyperCLOVA X SEED Think (32B)	–	62.9%	–	–
82	Ring-flash-2.0	–	62.8%	–	–
83	Olmo 3 7B Think	–	61.7%	–	–
84	Solar Pro 2 (Reasoning)	high	61.6%	–	–
85	Haiku 4.5	–	61.5%	–	–
86	GLM 4.5 V	–	60.4%	–	–
87	Nova 2.0 Omni (Low)	low	59.2%	–	–
88	Ling-flash-2.0	none	58.9%	–	–
89	Qwen 3 Coder	none	58.5%	–	–
90	o1 mini	–	57.6%	–	–
91	Mi:dm K 2.5 Pro Preview	–	57.6%	–	–
92	Qwen 3 32B	–	54.6%	–	–
93	K2-V2 (Medium)	medium	54.1%	–	–
94	Magistral Medium 1	medium	52.7%	–	–
95	Qwen 3 14B	–	52.3%	–	–
96	Exaone 4.0 1.2B (Reasoning)	–	51.6%	–	–
97	Magistral Small 1	–	51.4%	–	–
98	GPT-4.1 Mini	none	48.3%	–	–
99	Sonnet 3.7	–	47.3%	–	–
100	Nova 2.0 Pro Preview (Non-reasoning)	none	47.3%	–	–
101	EXAONE 4.0 32B (Non-reasoning)	none	47.2%	–	–
102	Nova 2.0 Lite (Low)	low	46.9%	–	–
103	ERNIE 4.5 300B A47B	none	46.7%	–	–
104	Mistral Large 3	none	46.5%	–	–
105	Solar Pro 2 (Preview) (Reasoning)	high	46.2%	–	–
106	GPT-4.1	none	45.7%	–	–
107	Devstral 2	none	44.8%	–	–
108	Reka Flash 3	–	43.5%	–	–
109	Ling-mini-2.0	none	42.9%	–	–
110	Grok 3	none	42.5%	–	–
111	GPT-4o	none	42.5%	–	–
112	Solar Pro 2 (Non-reasoning)	none	42.4%	–	–
113	GLM 4.6 V	none	41.1%	–	–
114	Mistral Medium 3.1	none	40.6%	–	–
115	Qwen 3 8B	–	40.6%	–	–
116	DeepSeek V3	none	40.5%	–	–
117	Mistral Medium 3	none	40.0%	–	–
118	Llama 4 Maverick	none	39.7%	–	–
119	K2-V2 (Low)	low	39.3%	–	–
120	Solar Pro 2 (Preview) (Non-reasoning)	none	38.5%	–	–
121	Sonnet 3.5	none	38.1%	–	–
122	Kimi Linear 48B A3B Instruct	none	37.8%	–	–
123	NVIDIA Nemotron 3 Nano 30B A3B (Non-reasoning)	none	36.0%	–	–
124	Qwen 2.5 Max	none	35.9%	–	–
125	Ministral 3 14B	none	35.1%	–	–
126	Devstral Small 2	none	34.8%	–	–
127	Gemini 2.0 Pro	none	34.7%	–	–
128	Nova 2.0 Lite (Non-reasoning)	none	34.6%	–	–
129	NVIDIA Nemotron Nano 12B v2 VL (Non-reasoning)	none	34.5%	–	–
130	Devstral Medium	none	33.7%	–	–
131	QwQ 32B-Preview	–	33.7%	–	–
132	Gemini 2.0 Flash	none	33.4%	–	–
133	GPT-4.1 Nano	none	32.6%	–	–
134	Nova Premier	none	31.7%	–	–
135	Gemini 1.5 Pro	none	31.6%	–	–
136	Haiku 3.5	none	31.4%	–	–
137	Qwen 3 1.7B	–	30.8%	–	–
138	Nova 2.0 Omni (Non-reasoning)	none	30.5%	–	–
139	Ministral 3 8B	none	30.3%	–	–
140	Llama 4 Scout	none	29.9%	–	–
141	Sonar	none	29.5%	–	–
142	Qwen 2.5 Coder	none	29.5%	–	–
143	Sarvam M (Reasoning, Based on Mistral Small 3.1)	high	29.5%	–	–
144	Mistral Large 2	none	29.3%	–	–
145	Exaone 4.0 1.2B (Non-reasoning)	none	29.3%	–	–
146	GPT-4	none	29.1%	–	–
147	Llama Nemotron Super 49B v1.5 (Non-reasoning)	none	29.0%	–	–
148	Llama 3.3 70b	none	28.8%	–	–
149	Command A	none	28.7%	–	–
150	Llama 3.3	none	28.0%	–	–
151	Opus 3	none	27.9%	–	–
152	Qwen 2.5 72B	none	27.6%	–	–
153	Sonar Pro	none	27.5%	–	–
154	Mistral Small 3.2	none	27.5%	–	–
155	Gemini 1.5 Flash	none	27.3%	–	–
156	Grok 2	none	26.7%	–	–
157	Olmo 3 7B Instruct	none	26.6%	–	–
158	Pixtral Large	none	26.1%	–	–
159	Devstral Small	none	25.8%	–	–
160	Mistral Small 3	none	25.2%	–	–
161	Granite 4.0 H Small	none	25.1%	–	–
162	Qwen 2.5 32B	none	24.8%	–	–
163	Ministral 3 3B	none	24.7%	–	–
164	Grok Beta	none	24.1%	–	–
165	GPT-4o mini	none	23.4%	–	–
166	Nova Pro	none	23.3%	–	–
167	Phi-4	none	23.1%	–	–
168	Llama 3.2 90b	none	21.4%	–	–
169	Mistral Small 3.1	none	21.2%	–	–
170	Jamba Reasoning 3B	–	21.0%	–	–
171	Llama 3 70b	none	19.8%	–	–
172	Mistral	none	19.5%	–	–
173	Claude 2.1	none	19.5%	–	–
174	Gemini 2.0 Flash-Lite	none	18.5%	–	–
175	Jamba 1.7 Large	none	18.1%	–	–
176	Granite 4.0 Micro	none	18.0%	–	–
177	Mistral Large	none	17.8%	–	–
178	Sonnet 3	none	17.5%	–	–
179	Jamba 1.6 Large	none	17.2%	–	–
180	Claude 2.0	none	17.1%	–	–
181	Nova Lite	none	16.7%	–	–
182	Qwen 2.5 Turbo	none	16.3%	–	–
183	Qwen 2 72B	none	15.9%	–	–
184	DeepSeek Coder V2 Lite Instruct	none	15.8%	–	–
185	Haiku 3	none	15.4%	–	–
186	LFM2 8B A1B	none	15.1%	–	–
187	Mixtral 8x22B Instruct	none	14.8%	–	–
188	Gemma 3n E4B Instruct	none	14.6%	–	–
189	Jamba 1.5 Large	none	14.3%	–	–
190	Mistral Small	none	14.1%	–	–
191	Nova Micro	none	14.0%	–	–
192	Gemma 3n E4B Instruct Preview (May '25)	none	13.8%	–	–
193	Gemma 3 27B	none	13.7%	–	–
194	Gemma 3 12B	none	13.7%	–	–
195	Phi-4 Multimodal Instruct	none	13.1%	–	–
196	Granite 3.3 8B (Non-reasoning)	none	12.7%	–	–
197	Phi-4 Mini Instruct	none	12.6%	–	–
198	Command-R+ (Apr '24)	none	12.2%	–	–
199	Qwen 3 0.6B	–	12.1%	–	–
200	Llama 3.1 8b	none	11.6%	–	–
201	Phi-3 Mini Instruct 3.8B	none	11.6%	–	–
202	Gemini 1.0 Pro	none	11.6%	–	–
203	OpenChat 3.5 (1210)	none	11.5%	–	–
204	Granite 4.0 H 1B	none	11.5%	–	–
205	Gemma 3 4B	none	11.2%	–	–
206	Llama 3.2 11b	none	11.0%	–	–
207	Claude Instant	none	10.9%	–	–
208	Mistral Medium	none	9.9%	–	–
209	Llama 2 70b	none	9.8%	–	–
210	Llama 2 13b	none	9.8%	–	–
211	LFM 40B	none	9.6%	–	–
212	Llama 3 8b	none	9.6%	–	–
213	Gemma 3n E2B Instruct	none	9.5%	–	–
214	DBRX Instruct	none	9.3%	–	–
215	Llama 3.2 3b	none	8.3%	–	–
216	LFM2 2.6B	none	8.1%	–	–
217	Jamba 1.6 Mini	none	7.1%	–	–
218	OLMo 2 32B	none	6.8%	–	–
219	Mixtral 8x7B Instruct	none	6.6%	–	–
220	Jamba 1.5 Mini	none	6.2%	–	–
221	Jamba 1.7 Mini	none	6.1%	–	–
222	Command-R (Mar '24)	none	4.8%	–	–
223	Granite 4.0 1B	none	4.7%	–	–
224	OLMo 2 7B	none	4.1%	–	–
225	Molmo 7B-D	none	3.9%	–	–
226	Granite 4.0 350M	none	2.4%	–	–
227	LFM2 1.2B	none	2.0%	–	–
228	Llama 3.2 1b	none	1.9%	–	–
229	Granite 4.0 H 350M	none	1.9%	–	–
230	Gemma 3 1B	none	1.7%	–	–
231	Gemma 3 270M	none	0.3%	–	–
232	Llama 2 7b	none	0.2%	–	–