Claude Opus 4.8
Anthropic
Parameters
—
Architecture
—
Released
—
License
—
About
No detailed description available for this model.
Benchmark Scores
| Benchmark | Score | Date | Standing |
|---|---|---|---|
|
AgentWorldBench MCP
general_agent
|
54.93
|
24.06.2026 |
#14/14
better than 0%
|
|
AgentWorldBench Search
general_agent
|
83.12%
|
24.06.2026 |
#4/14
better than 77%
|
|
AgentWorldBench Terminal
general_agent
|
100.00%
|
24.06.2026 |
#1/14
better than 100%
|
|
AgentWorldBench SWE
general_agent
|
85.86%
|
24.06.2026 |
#7/14
better than 54%
|
|
AgentWorldBench Android
general_agent
|
97.43%
|
24.06.2026 |
#2/14
better than 92%
|
|
AgentWorldBench Web
general_agent
|
100.00%
|
24.06.2026 |
#1/14
better than 100%
|
|
AgentWorldBench OS
general_agent
|
74.30%
|
24.06.2026 |
#5/14
better than 69%
|
|
AgentWorldBench Overall
general_agent
|
83.16%
|
24.06.2026 |
#4/14
better than 77%
|
|
Terminal Bench 2.1
coding_agent
|
93.35%
|
18.08.2026 |
#10/44
better than 77%
|
|
SWE-bench Pro
coding_agent
|
77.15%
|
18.08.2026 |
#4/77
better than 96%
|
|
NL2Repo-Bench
coding_agent
|
89.01%
|
18.08.2026 |
#2/23
better than 95%
|
|
FrontierSWE
coding_agent
|
69.26%
|
18.08.2026 |
#7/12
better than 45%
|
|
MLS-Bench-Lite
coding_agent
|
69.40%
|
18.08.2026 |
#4/9
better than 62%
|
|
PaperBench
coding_agent
|
79.65%
|
18.08.2026 |
#4/10
better than 67%
|
|
AndroidBench
coding_agent
|
47.50%
|
18.08.2026 |
#4/5
better than 25%
|
|
QwenSWEBench
coding_agent
|
93.78%
|
18.08.2026 |
#2/9
better than 88%
|
|
QwenQoderBench
coding_agent
|
98.48%
|
18.08.2026 |
#2/5
better than 75%
|
|
QwenReactBench
coding_agent
|
67.24%
|
18.08.2026 |
#3/5
better than 50%
|
|
QwenSVGBench
coding_agent
|
57.53%
|
18.08.2026 |
#4/5
better than 25%
|
|
CoWorkBench
general_agent
|
88.31%
|
18.08.2026 |
#4/12
better than 73%
|
|
WorkSpaceBench
general_agent
|
51.19%
|
18.08.2026 |
#5/11
better than 60%
|
|
JobBench
general_agent
|
57.58%
|
18.08.2026 |
#10/19
better than 50%
|
|
SkillsBench
general_agent
|
79.56%
|
18.08.2026 |
#4/9
better than 62%
|
|
Agents' Last Exam
general_agent
|
50.31%
|
18.08.2026 |
#10/29
better than 61%
|
|
Toolathlon Verified
general_agent
|
96.61%
|
18.08.2026 |
#4/35
better than 88%
|
|
WideSearch
general_agent
|
73.78%
|
18.08.2026 |
#13/24
better than 48%
|
|
GPQA Diamond
stem_reasoning
|
95.65%
|
18.08.2026 |
#14/80
better than 84%
|
|
Humanity's Last Exam
stem_reasoning
|
82.58%
|
18.08.2026 |
#6/71
better than 93%
|
|
IFBench
instruction_following
|
64.23%
|
18.08.2026 |
#34/42
better than 20%
|
|
$OneMillion-Bench
general_capabilities
|
41.80
|
18.08.2026 |
#11/11
better than 0%
|
|
HealthBench
general_capabilities
|
55.93%
|
18.08.2026 |
#5/6
better than 20%
|
|
PLawBench
general_capabilities
|
74.83%
|
18.08.2026 |
#4/5
better than 25%
|
|
PRBench-Legal
general_capabilities
|
46.15%
|
18.08.2026 |
#4/5
better than 25%
|
|
PRBench-Finance
general_capabilities
|
44.35%
|
18.08.2026 |
#4/5
better than 25%
|
|
MRCR v2 256K (8-needle)
long_context
|
83.20
|
18.08.2026 |
#4/4
better than 0%
|
|
LongBench v2
long_context
|
100.00%
|
18.08.2026 |
#1/22
better than 100%
|
|
CritPt (no tools)
stem_reasoning
|
64.04%
|
17.06.2026 |
#5/21
better than 75%
|
|
AIME 26
stem_reasoning
|
95.54%
|
17.06.2026 |
#15/45
better than 68%
|
|
HMMT Nov 25
stem_reasoning
|
100.00%
|
17.06.2026 |
#1/20
better than 95%
|
|
HMMT Feb 26
stem_reasoning
|
98.70%
|
17.06.2026 |
#3/44
better than 93%
|
|
IMOAnswerBench
stem_reasoning
|
86.51%
|
17.06.2026 |
#11/30
better than 66%
|
|
NL2Repo
coding_agent
|
91.38%
|
17.06.2026 |
#2/41
better than 95%
|
|
ProgramBench
coding_agent
|
100.00%
|
17.06.2026 |
#1/18
better than 100%
|
|
Terminal-Bench 2.1 (Terminus-2)
coding_agent
|
94.40%
|
17.06.2026 |
#8/44
better than 84%
|
|
Terminal-Bench 2.1 (Best Reported Harness)
coding_agent
|
68.75%
|
17.06.2026 |
#3/5
better than 50%
|
|
PostTrainBench
coding_agent
|
82.25%
|
17.06.2026 |
#6/22
better than 76%
|
|
Tool Decathlon
general_agent
|
100.00%
|
17.06.2026 |
#1/11
better than 100%
|
|
HLE (with tools)
stem_reasoning
|
78.43%
|
18.08.2026 |
#11/49
better than 77%
|
|
Program Bench
coding_agent
|
89.42%
|
23.08.2026 |
#2/7
better than 83%
|
|
Kimi Claw 24/7 Bench
agentic
|
75.76%
|
23.08.2026 |
#2/4
better than 67%
|
|
MCPMark-Verified
agentic
|
17.91%
|
23.08.2026 |
#3/4
better than 33%
|
|
SWE-Marathon
coding_agent
|
49.31%
|
17.06.2026 |
#11/17
better than 38%
|
|
Kimi Code Bench V2
coding_agent
|
62.98%
|
23.08.2026 |
#4/7
better than 50%
|
|
GDPVal-AA v2
general_agent
|
86.40%
|
27.08.2026 |
#14/38
better than 65%
|
|
MCP-Atlas
general_agent
|
85.12%
|
17.06.2026 |
#9/54
better than 85%
|
|
DeepSWE 1.1
coding_agent
|
76.56%
|
18.08.2026 |
#17/34
better than 52%
|
|
Automation-Bench
general_agent
|
31.54%
|
18.08.2026 |
#19/27
better than 27%
|
What is a good AgentWorldBench MCP score? What is a good AgentWorldBench Search score? What is a good AgentWorldBench Terminal score?
No comments yet
Be the first to share your thoughts!