Coding

Vals LiveCodeBench leaderboard

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

As of 2026-10-07, the highest measured score on Vals LiveCodeBench is 90.5% by Claude Fable 5.1. 94 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Ember-1	91.8%	estimated ± 9.9 pp, low confidence
2	Claude Fable 5.1	90.5%	measured
3	Claude Fable 5	89.8%	measured
4	Gemini 3.8 Flash	89.5%	measured
5	Composer 2	89.4%	estimated ± 3.8 pp, low confidence
6	Claude Opus 5.5	89.3%	estimated ± 1.9 pp, medium confidence
7	Composer 2 Fast	89.1%	estimated ± 3.8 pp, low confidence
8	Claude Opus 5	89.0%	measured
9	GPT-6 Astra	89.0%	estimated ± 1.9 pp, high confidence
10	Sakana Fugu	88.8%	estimated ± 0.8 pp, low confidence
11	Gemini 3.7 Flash	88.7%	measured
12	Grok 4.7	88.6%	estimated ± 2.7 pp, low confidence
13	Step 5 Preview	88.5%	estimated ± 0.8 pp, low confidence
14	Gemini 3.1 Pro	88.5%	measured
15	Claude Mythos 5	88.5%	estimated ± 4.7 pp, medium confidence
16	Sakana Fugu-Ultra	88.5%	estimated ± 0.8 pp, low confidence
17	Grok 4.6	88.2%	measured
18	Gemini 3.6 Flash	88.1%	measured
19	SWE-2	88.1%	estimated ± 1.9 pp, high confidence
20	GPT-5.2-Codex	88.0%	measured
21	Qwen3.8 Max	87.9%	measured
22	Claude Opus 4.8	87.8%	measured
23	Gemini 3.5 Flash	87.6%	measured
24	DeepSeek V4 Pro 0813	87.5%	measured
25	Grok 4.5	87.4%	measured
26	DeepSeek V4 Flash 0731	87.3%	measured
27	GPT-5.3 Codex	87.3%	measured
28	Kimi K3	87.2%	measured
29	Qwen3.7 Max	87.1%	measured
30	Claude Haiku 5.5	87.1%	estimated ± 1.9 pp, high confidence
31	Claude Sonnet 5.5	87.0%	estimated ± 1.9 pp, high confidence
32	Kimi K2.6	86.8%	measured
33	Muse Spark 1.3	86.8%	estimated ± 2.7 pp, low confidence
34	Hy4 preview	86.7%	estimated ± 4.7 pp, medium confidence
35	Qwen3.7 Plus	86.6%	estimated ± 0.8 pp, medium confidence
36	Ornith-1.5-397B	86.5%	estimated ± 4.7 pp, medium confidence
37	Claude Opus 4.7 (Adaptive)	86.3%	estimated ± 4.7 pp, medium confidence
38	Beam	86.2%	estimated ± 0.8 pp, medium confidence
39	GPT-5.4	86.2%	estimated ± 3.8 pp, low confidence
40	Qwen3.8-Omni-Flash	86.0%	estimated ± 4.7 pp, medium confidence
41	Nemotron 3 Ultra	86.0%	measured
42	Qwen3.6 Plus	86.0%	measured
43	Kimi K2.5	86.0%	estimated ± 0.8 pp, medium confidence
44	SWE-1.7	86.0%	estimated ± 1.9 pp, high confidence
45	GPT-5.6 Terra	85.9%	measured
46	Inkling-Small	85.9%	measured
47	Muse Spark 1.1	85.9%	measured
48	Qwen3.8-Flash-Next	85.8%	estimated ± 4.7 pp, medium confidence
49	Ornith-1.0-397B	85.7%	estimated ± 4.7 pp, medium confidence
50	Gemini 3 Flash	85.6%	measured
51	GPT-5.1-Codex	85.6%	measured
52	Qwen 3.6 Max (preview)	85.6%	estimated ± 0.8 pp, medium confidence
53	Inkling	85.5%	measured
54	GPT-5.5	85.3%	measured
55	dots3-note Preview	85.2%	estimated ± 4.7 pp, medium confidence
56	Claude Opus 4.7	85.1%	measured
57	Muse Glimmer 30B	84.7%	estimated ± 0.8 pp, medium confidence
58	Atria Dawn Preview	84.7%	estimated ± 4.7 pp, medium confidence
59	Ornith-1.5-35B-A3B	84.7%	estimated ± 4.7 pp, medium confidence
60	Laguna S 2.1	84.6%	estimated ± 4.7 pp, medium confidence
61	Grok 4.3	84.5%	measured
62	Grok 4.20	84.3%	measured
63	Hy3 Preview	84.1%	estimated ± 0.8 pp, low confidence
64	A.X K2	84.1%	estimated ± 0.8 pp, low confidence
65	GPT-5.6 Luna	84.0%	estimated ± 2.7 pp, low confidence
66	GPT-5.4 nano	84.0%	measured
67	Ling 3.0 Flash	84.0%	measured
68	Qwen3.8-27B	84.0%	measured
69	Ling 3.0 Flash FP8	83.9%	estimated ± 0.8 pp, low confidence
70	GPT-5.1-Codex-Max	83.6%	measured
71	Granite 4.2 30B	83.5%	estimated ± 0.8 pp, low confidence
72	Claude Opus 4.5	83.4%	estimated ± 4.7 pp, medium confidence
73	Mercury 2.5	83.3%	estimated ± 0.8 pp, low confidence
74	Qwen3.5 Flash	83.3%	measured
75	K-EXAONE 2.0	83.2%	estimated ± 0.8 pp, low confidence
76	Step 3.7 Flash	82.9%	estimated ± 4.7 pp, medium confidence
77	Granite 4.2 8B	82.9%	estimated ± 0.8 pp, low confidence
78	GPT-5.6 Sol	82.6%	measured
79	Gemma 4 31B	82.6%	estimated ± 3.8 pp, low confidence
80	GPT-5.2	82.5%	estimated ± 4.7 pp, medium confidence
81	GLM-5	82.4%	estimated ± 3.8 pp, low confidence
82	Claude Sonnet 5	82.4%	measured
83	GLM-4.7	82.2%	measured
84	MiniMax M3	82.2%	measured
85	Claude Sonnet 4.6	82.1%	measured
86	Kimi K2.7 Code	82.1%	measured
87	Nemotron 3 Nano Omni 30B A3B	81.8%	estimated ± 0.8 pp, low confidence
88	Claude Opus 4.6	81.8%	estimated ± 1.9 pp, high confidence
89	Nemotron 3.5 Lightning 30B A3B NVFP4	81.7%	estimated ± 0.8 pp, low confidence
90	Grok 4	81.5%	estimated ± 3.8 pp, low confidence
91	GPT-5.4 mini	81.5%	measured
92	MiMo-V2.5	81.5%	measured
93	GLM-5.1	81.4%	measured
94	MiMo-V2.5-Pro	81.4%	measured
95	Agents-A1-4B	81.2%	estimated ± 0.8 pp, low confidence
96	GPT-OSS 120B	81.1%	estimated ± 3.8 pp, low confidence
97	DeepSeek V3.2	81.1%	estimated ± 3.8 pp, low confidence
98	GLM-4.6	81.0%	measured
99	GPT-OSS 20B	80.9%	estimated ± 3.8 pp, low confidence
100	Qwen3.6-27B	80.8%	estimated ± 4.7 pp, medium confidence
101	BTL-4	80.6%	estimated ± 9.9 pp, low confidence
102	Ling 2.6 Flash	80.6%	estimated ± 0.8 pp, low confidence
103	GLM-5.3	80.5%	measured
104	GLM-5.3-Flash	80.5%	measured
105	MiniCPM5-2B	80.4%	estimated ± 0.8 pp, low confidence
106	MAI-Thinking-1	80.2%	estimated ± 4.7 pp, medium confidence
107	MiMo-V2-Pro	80.1%	estimated ± 9.9 pp, low confidence
108	Gemini 3.1 Flash-Lite	80.1%	measured
109	MiniMax M2.7	79.9%	measured
110	Granite 4.2 3B	79.9%	estimated ± 0.8 pp, low confidence
111	Muse Spark	79.8%	estimated ± 4.7 pp, medium confidence
112	Apodex 1.1	79.8%	estimated ± 9.9 pp, low confidence
113	Mistral Medium 3.5 128B	79.6%	estimated ± 9.9 pp, low confidence
114	Claude Sonnet 4.5	79.1%	estimated ± 9.9 pp, low confidence
115	Gemini 3.5 Flash-Lite	79.0%	measured
116	Composer 2.5	78.7%	estimated ± 2.7 pp, low confidence
117	Kimi K2.5 (Reasoning)	78.6%	estimated ± 9.9 pp, low confidence
118	Qwen3.5 397B	78.3%	estimated ± 4.7 pp, medium confidence
119	Ornith-1.0-35B	77.7%	estimated ± 4.7 pp, medium confidence
120	DeepSeek V4.1 Flash	77.4%	estimated ± 6.1 pp, low confidence
121	MiMo-V2.6-Flash	77.4%	estimated ± 6.1 pp, low confidence
122	MiMo-V2.6-Pro	77.4%	estimated ± 6.1 pp, low confidence
123	Qwen3.6-35B-A3B	76.5%	estimated ± 4.7 pp, medium confidence
124	MiMo-V2-Omni	75.8%	estimated ± 9.9 pp, low confidence
125	Claude 4.1 Opus	75.3%	estimated ± 9.9 pp, low confidence
126	Laguna XS 2.1	73.7%	estimated ± 4.7 pp, medium confidence
127	MiMo-V2-Flash	73.6%	estimated ± 9.9 pp, low confidence
128	Ornith-1.5-9B	73.6%	estimated ± 4.7 pp, medium confidence
129	Claude 4 Sonnet	72.5%	estimated ± 9.9 pp, low confidence
130	MAI-Code-1.1-Flash	72.3%	estimated ± 9.9 pp, low confidence
131	Qwen3.5-27B	72.0%	estimated ± 9.9 pp, low confidence
132	Qwen3.5-122B-A10B	71.3%	estimated ± 9.9 pp, low confidence
133	GLM-5.2	69.5%	measured
134	Solar Pro 4	68.9%	estimated ± 9.9 pp, low confidence
135	Solar Open 2	68.5%	estimated ± 9.9 pp, low confidence
136	Laguna M.1	68.1%	measured
137	Laguna XS.2	67.8%	measured
138	GLM-4.5	67.4%	measured
139	Qwen3.5-35B-A3B	66.2%	estimated ± 9.9 pp, low confidence
140	Ornith-1.0-9B	64.1%	estimated ± 4.7 pp, low confidence
141	Grok Code Fast 1	62.0%	measured
142	LongCat-Flash-Lite-Sparse	57.8%	estimated ± 4.7 pp, low confidence
143	Gemini 2.5 Pro	54.6%	estimated ± 9.9 pp, low confidence
144	Ternary Bonsai 2 27B	47.4%	estimated ± 9.9 pp, low confidence
145	Claude Haiku 4.5	41.2%	measured
146	GPT-4.1	32.0%	estimated ± 9.9 pp, low confidence
147	ZAYA1-74B-Preview	28.7%	estimated ± 9.9 pp, low confidence
148	LLaDA2.2-flash	20.8%	estimated ± 4.7 pp, low confidence
149	o3-mini	20.3%	estimated ± 9.9 pp, low confidence
150	Claude 3.5 Sonnet	19.8%	estimated ± 9.9 pp, low confidence
151	DeepSeek V3	8.9%	estimated ± 9.9 pp, low confidence
152	GPT-4.1 mini	0.3%	estimated ± 9.9 pp, low confidence

Vals LiveCodeBench leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General