Claude 4.5 Opus
Anthropic
Parameters
—
Architecture
—
Released
—
License
—
About
No detailed description available for this model.
Benchmark Scores
| Benchmark | Score | Date | Standing |
|---|---|---|---|
|
SWE-bench Verified
coding_agent
|
92.32%
|
22.04.2026 |
#4/60
better than 95%
|
|
SWE-bench Multilingual
coding_agent
|
85.80%
|
22.04.2026 |
#8/48
better than 81%
|
|
Terminal-Bench 2.0
coding_agent
|
74.86%
|
22.04.2026 |
#8/23
better than 64%
|
|
SkillsBench Avg5
coding_agent
|
65.65%
|
22.04.2026 |
#10/16
better than 40%
|
|
QwenWebBench
coding_agent
|
100.00%
|
22.04.2026 |
#1/8
better than 100%
|
|
NL2Repo
coding_agent
|
50.62%
|
22.04.2026 |
#20/41
better than 52%
|
|
Claw-Eval Avg
coding_agent
|
92.72%
|
22.04.2026 |
#3/40
better than 95%
|
|
QwenClawBench
coding_agent
|
97.75%
|
22.04.2026 |
#3/17
better than 88%
|
|
MMLU-Pro
knowledge
|
95.16%
|
22.04.2026 |
#2/48
better than 98%
|
|
MMLU-Redux
knowledge
|
100.00%
|
22.04.2026 |
#1/30
better than 100%
|
|
SuperGPQA
knowledge
|
100.00%
|
22.04.2026 |
#1/28
better than 100%
|
|
GPQA Diamond
stem_reasoning
|
86.20%
|
22.04.2026 |
#40/80
better than 49%
|
|
Humanity's Last Exam
stem_reasoning
|
54.36%
|
22.04.2026 |
#32/71
better than 54%
|
|
LiveCodeBench v6
stem_reasoning
|
88.13%
|
22.04.2026 |
#19/49
better than 62%
|
|
HMMT Feb 25
stem_reasoning
|
95.78%
|
22.04.2026 |
#3/17
better than 88%
|
|
HMMT Nov 25
stem_reasoning
|
83.76%
|
22.04.2026 |
#8/20
better than 63%
|
|
HMMT Feb 26
stem_reasoning
|
83.94%
|
22.04.2026 |
#20/44
better than 56%
|
|
IMOAnswerBench
stem_reasoning
|
87.22%
|
22.04.2026 |
#10/30
better than 69%
|
|
AIME 26
stem_reasoning
|
94.77%
|
22.04.2026 |
#19/45
better than 59%
|
|
MMMU
vision_language
|
90.61%
|
22.04.2026 |
#7/19
better than 67%
|
|
DynaMath
vision_language
|
10.11%
|
22.04.2026 |
#11/14
better than 23%
|
|
MMStar
vision_language
|
62.32%
|
22.04.2026 |
#14/16
better than 13%
|
|
CharXiv (RQ)
document_understanding
|
9.88%
|
22.04.2026 |
#28/34
better than 18%
|
|
CC-OCR
document_understanding
|
63.31%
|
22.04.2026 |
#9/14
better than 38%
|
|
CountBench
spatial_intelligence
|
7.69%
|
22.04.2026 |
#12/13
better than 8%
|
|
V-Star
vision_language
|
21.11%
|
22.04.2026 |
#18/20
better than 5%
|
|
MMMU-Pro
vision_language
|
76.60%
|
22.04.2026 |
#23/35
better than 35%
|
|
C-Eval
knowledge
|
91.74%
|
22.04.2026 |
#3/15
better than 86%
|
|
VideoMME (w sub.)
video_understanding
|
41.43%
|
22.04.2026 |
#16/17
better than 6%
|
|
SimpleVQA
vision_language
|
78.82%
|
22.04.2026 |
#3/16
better than 87%
|
|
MLVU
video_understanding
|
65.00%
|
22.04.2026 |
#11/13
better than 17%
|
|
ERQA
spatial_intelligence
|
24.56%
|
22.04.2026 |
#16/21
better than 25%
|
|
SWE-bench Pro
coding_agent
|
63.66%
|
22.04.2026 |
#30/77
better than 62%
|
|
Claw-Eval Pass^3
coding_agent
|
64.92%
|
22.04.2026 |
#7/21
better than 70%
|
|
VideoMMMU
video_understanding
|
90.86%
|
22.04.2026 |
#5/19
better than 72%
|
|
MVBench
video_understanding
|
46.86%
|
22.04.2026 |
#11/12
better than 9%
|
What is a good SWE-bench Verified score? What is a good SWE-bench Multilingual score? What is a good Terminal-Bench 2.0 score?
No comments yet
Be the first to share your thoughts!