Inference
EN
LiveBench

LiveBench Reasoning

Short logic, spatial and theory-of-mind questions with one correct answer. Questions from the 2026-06-25 release; LiveBench 2.0 keeps them.

ModelsModels with a result.80
Top resultBest result on this benchmark.95.1%Gemini 4 Argon
Top-3 spreadPoints from first to third.2.5 pts
UpdatedDate the source changed the results.14 Oct 2026

Result and cost

78 of 80 results

Results

80 results
#ModelResultThe score from the source.CostUS dollars to run one task.
01Gemini 4 Argonhigh effort95.1%$0.078
02GPT-6 Astramax effort92.7%$0.39
03GPT-6.1 Solmax effort92.6%$0.08
04Claude Opus 5.5max effort92.2%$0.252
05Claude Fable 5.1max effort91.7%$0.46
05GPT-5.6 Solmax effort91.7%$0.285
07Claude Sonnet 5.5max effort91.6%$0.209
08Claude Opus 5max effort91.2%$0.168
08Grok 4.6xhigh effort91.2%$0.087
10Kimi K390.7%$0.195
11Muse Wondermax effort90.6%$0.298
11GPT-5.6 Terramax effort90.6%$0.234
13Smaug Agentic90.3%$0.189
14Muse Spark 1.2xhigh effort90.0%$0.099
15Claude Fable 5max effort89.7%$0.602
15Muse Spark 1.3xhigh effort89.7%$0.136
15GPT-5.5xhigh effort89.7%$0.289
18Gemini 3.8 Flashhigh effort89.3%$0.148
19Claude Opus 4.8max effort89.2%$0.606
20GPT-6.1 Terramax effort88.7%$0.04
20Claude Sonnet 5xhigh effort88.7%$0.383
20Claude Opus 4.6 Thinking Autohigh effort88.7%$0.277
20GPT-6 Solmax effort88.7%$0.121
24Qwen3.8 Max88.2%$0.131
25GPT-5.4xhigh effort88.1%$0.213
26Gemini 3.7 Flashhigh effort87.8%$0.069
27Muse Spark 1.1xhigh effort87.7%$0.075
28Qwen3.8-Flash-Next87.4%$0.022
29Claude Opus 4.7xhigh effort87.2%$0.303
29Grok 4.587.2%$0.088
31DeepSeek V4.1 Flash 1017max effort86.7%$0.023
32DeepSeek V4 Flash 073186.6%$0.0063
33Claude Sonnet 4.6max effort86.4%$0.313
34Smaug Flash86.2%$0.0086
35DeepSeek V4 Pro 081385.8%$0.037
35GLM-5.385.8%$0.265
37GPT-5.6 Lunamax effort85.6%$0.138
38DeepSeek V4 Flash Vision Exp85.4%$0.042
39Mistral Large 4.0high effort85.2%$0.054
40Gemini 3.6 Flashhigh effort85.1%$0.097
41Claude Sonnet 4.6 Thinking Automedium effort84.8%$0.207
42Gemini 3.1 Pro Previewhigh effort84.0%$0.148
43Mistral Large 4high effort83.9%$0.061
44Step 5 Previewmax effort83.8%$0.048
45Claude Haiku 5.5max effort83.5%$0.021
46Qwen3.7 Max83.3%$0.102
47GPT-5.2high effort83.2%$0.114
48MiniMax M3.1 Flashmax effort82.9%$0.011
49Kimi K2.7 Code82.8%$0.053
49Smaug Mini82.8%$0.019
51DeepSeek V4 Pro 042382.7%$0.041
51Grok 4.7xhigh effort82.7%$0.385
53Gemini 3.5 Flashhigh effort82.0%$0.149
54GPT-6 Lunamax effort81.8%$0.021
55GPT-5.4 nanoxhigh effort81.1%$0.044
56Beam 1.0max effort80.2%$0.04
57Claude Opus 4.5 Thinking 64Khigh effort80.1%$0.499
58Qwen3.8-27B80.0%$0.016
59MiMo-V2.6-Prohigh effort79.6%$0.024
60Kimi K2.6 Thinking79.4%$0.089
61Nemotron Super VL-3.5max effort78.8%$0.013
62GLM-5.278.6%$0.121
63Inklingxhigh effort78.3%$0.121
64GPT-5.2-Codex77.7%$0.169
65GLM-5.3-Flash77.6%$0.014
66Ox Alpha Max76.6%$0
67Grok Build 0.176.4%$0.0021
68Qwen3.6 Plus75.8%$0.094
69Smaug Nemotron 3.5 Nvfp4max effort75.1%$0.013
70MiniMax M374.5%$0.019
71Muse Spark Openmax effort73.7%$0.043
72Nemotron Super VL-3.5 Nvfp4max effort71.7%$0.012
73GPT-5.4 minixhigh effort71.3%$0.157
74Grok 4.370.8%$0.041
75DeepSeek V4 Flash 042370.6%$0.014
76Qwen3.6-27B70.3%$0.133
77Gemma 4 31Bmax effort68.3%$0.0038
78Claude Haiku 4.5max effort61.7%—
79Gemini 3.5 Flash-Litehigh effort60.2%$0.031
80Nemotron 3.5 Lightning 30B A3B42.9%$0.022