어제자 챗gpt에 묻힌 구글의 개쩌는 ai모델

Gemini 2.5 Pro

Gemini
OpenAl
UDErVA
Cllaude
Grok
Deepseek
25 Pro
03-mini
GPT-4,5
3.7 Sonnet
Beta
[anchrrrk
Eapztnizr:al ‘DJ-251
EtciJcd Irung
Lt undzu Taiilr]
767r7-
know cEqc
Hurvanity
Last Exam
18.896
14.096*
6.490
8.996
8.696*
(no tools)
5clonEa
anile
GPOA diamond
eatiia
84.096
79.706
71.496
78.206
80.286
71.596
rrJtirlc JcTIIE:-
84.89
84.693
M il Mihilrs
AlME 2025
paterat
86.7%6
86,596
49,596
77,396
70.096
rritirlo arttcTr:도
93.396
Meilurril
grala
AIMIL 2029
‘Fratjif
92.0%
87,396
36.796
61,396
83,996
79,896
Irallildle dt-iti (
80.096
93.396
느디JU qEr urutizn
LiveCodeBerch v5
FIretr끼il
70.495
74.196
70.696
64.395
하ra IUe – n
79.4%6
CUJ0 cJitirg
Aider Polyglot
74.086
60.496
44.996
64.996
56.996
68.696
-다C ‘ 7ie
Adenitrcoyllnu
SWE-bench verified
63.896
49.396
38.096
70.3%6
49.,206
FaGluabl;
SimpleOA
52,996
13,836
62.596
43.696
30,196
rEABorwfi
MMMU
angltitisraf
81.7%6
Iu MM – MUT
74,496
75,096
76.096
‘MM sup0c1
miattom
nu MM surou
78.096
IO VM Guprci
MTVa Jt undurytndinn
Eval (Reka)
69.496
n MV SuSI
IO VM GuXI
Lani corturt
MRCR
91.596
36.396
48,892
83,1%6
MJnilitii 기
Global MMLU (Lite)
89,896
VDe

Global
Model
Organization
Average
gemini-2.5-
Google
82.35
pro-exp-03-25
claude-3-7-
Anthropic
76.10
sonnet-
thinking
03-mini-2025-
OpenAl
75.88
01-37-high
01-2024-12-
OpenAl
75.67
17-high
9w9-32b
Alibaba
71.96
deepseek-r7
DeepSeek
71.57
03-mini-2025-
OpenAl
70.01
01-31-
medium
OpenAl
68.95
preview
gemini-2.0-
Google
66.92
flash-
thinking-exp-
01-21
gpt-4.5-