Skip to content

Benchmark

LiveBench AMPS_Hard

Scores are percentages.

Models measured
51
Scores recorded
51
Sources
1
IQ weight
1.02

Score distribution

95.5%
Mean
98%
Median
6.6%
Std dev
89%
10th percentile
99%
90th percentile
66%
Min
66%99%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.02 for chat models, measured over 51 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Claude Opus 4.564k, high· LiveBench data99%
2Claude Opus 5.5xhigh· LiveBench data99%
3Gemini 3.8 Flashhigh· LiveBench data99%
4GLM 5.3· LiveBench data99%
5Grok 4.5· LiveBench data99%
6Inkling FP4xhigh· LiveBench data99%
7Muse Spark 1.3xhigh· LiveBench data99%
8Qwen 3.8 Flash Next· LiveBench data99%
9Claude Haiku 5.5xhigh· LiveBench data98%
10Claude Opus 4.7xhigh· LiveBench data98%
11Claude Sonnet 5xhigh· LiveBench data98%
12Claude Sonnet 5.5xhigh· LiveBench data98%
13DeepSeek V4 Flash· LiveBench data98%
14DeepSeek V4 Pro· LiveBench data98%
15DeepSeek V4 Pro 0813· LiveBench data98%
16Gemini 3.1 Pro Preview· LiveBench data98%
17Gemini 3.5 Flashhigh· LiveBench data98%
18Gemini 3.6 Flashhigh· LiveBench data98%
19Gemini 3.7 Flashhigh· LiveBench data98%
20GLM 5.2· LiveBench data98%
21GPT 5.4xhigh· LiveBench data98%
22GPT 5.4 Nanoxhigh· LiveBench data98%
23GPT-5.2 Codex· LiveBench data98%
24GPT-6.1 Solxhigh· LiveBench data98%
25Grok 4.7xhigh· LiveBench data98%
26Mistral Large 4high· LiveBench data98%
27Qwen 3.8 27B· LiveBench data98%
28Qwen3.7 Max· LiveBench data98%
29Qwen3.8 Max· LiveBench data98%
30Claude Opus 4.6auto, high· LiveBench data97%
31DeepSeek V4 Flash 0731· LiveBench data97%
32GLM 5.3 Flash· LiveBench data97%
33GPT 5.2high· LiveBench data97%
34GPT 5.5xhigh· LiveBench data97%
35GPT-5.4 minixhigh· LiveBench data97%
36Grok 4.3· LiveBench data97%
37Grok 4.6· LiveBench data97%
38Kimi K2.6thinking· LiveBench data97%
39Kimi K2.7 Code· LiveBench data97%
40Kimi K3· LiveBench data97%
41Nemotron 3.5 Lightning 30B A3B· LiveBench data97%
42Qwen3.6 Plus· LiveBench data97%
43Nemotron 3 Ultra· LiveBench data96%
44Gemini 3.5 Flash Litehigh· LiveBench data95%
45Qwen3.6 27B· LiveBench data93%
46Muse Spark 1.2xhigh· LiveBench data89%
47DeepSeek V4 Flash Vision Exp· LiveBench data88%
48Muse Spark 1.1xhigh· LiveBench data82%
49Claude Sonnet 4.6auto, medium· LiveBench data76%
50MiniMax M3· LiveBench data75%

Showing 1–50 of 51 results.