🧠 Reasoning · llm-stats.com

GPQA Diamond

Graduate-level science questions in physics, chemistry and biology, designed to be Google-proof.

45
نماذج مختبرة
Claude Mythos Preview
النموذج الأعلى
94.6%
أفضل درجة
Anthropic
المزوّد
#النموذجGPQA DiamondالمزوّدPodium Score
1
94.6%
Anthropic97.4
2
94.6%
OpenAI80.8
3
94.3%
Google60.6
4
94.2%
Anthropic52.0
5
93.6%
Anthropic76.0
6
GPT-5.5
OpenAI
93.6%
OpenAI54.0
7
Kimi K3
Moonshot AI
93.5%
Moonshot AI83.3
8
Claude Opus 5
Anthropic
93.2%
Anthropic81.4
9
93%
xAI61.3
10
92.9%
OpenAI66.4
11
MiniMax M3
MiniMax
92.9%
MiniMax39.6
12
92.8%
Google65.9
13
GPT-5.4
OpenAI
92.8%
OpenAI51.0
14
92.6%
Anthropic93.4
15
92.6%
Alibaba79.8
16
92.4%
Alibaba51.0
17
GPT-5.2
OpenAI
92.4%
OpenAI49.2
18
92.3%
OpenAI49.1
19
92.2%
Google50.0
20
91.9%
Google51.9
21
91.5%
OpenAI52.4
22
91.3%
Anthropic61.2
23
91.2%
Z.ai59.1
24
Kimi K2.6
Moonshot AI
90.5%
Moonshot AI49.0
25
90.4%
Google42.4
26
90.4%
Tencent38.9
27
90.4%
Alibaba32.4
28
90.3%
Alibaba42.2
29
90.1%
DeepSeek46.0
30
89.9%
Anthropic38.1
31
89.8%
Meta69.9
32
Kimi K2.7 Code
Moonshot AI
89.6%
Moonshot AI41.8
33
89.5%
Meta47.0
34
Seed 2.0 Pro
ByteDance
88.9%
ByteDance40.0
35
88.4%
xAI54.5
36
88.4%
Alibaba23.9
37
GPT-5.1
OpenAI
88.1%
OpenAI42.3
38
88.1%
DeepSeek42.0
39
Inkling
Thinking Machines
87.2%
Thinking Machines31.7
40
87%
Anthropic52.1
41
87%
NVIDIA19.0
42
86.6%
Xiaomi48.0
43
86.6%
Alibaba22.0
44
GPT-5
OpenAI
85.7%
OpenAI23.0
45
80%
Anthropic45.4
بيانات الاختبار مجمّعة من llm-stats.com. اعرف المزيد في المنهجية.