SShogo Bench

Claude-Opus-4

Flagship Claude model for deep reasoning, coding, and long-horizon agents

Shogo Agent Index

Supported#75 of 92
27.9range 20.4–35.4 · 32% of weight measured
Agentic coding
29
34% measured
Long-horizon autonomy
38
50% measured
Knowledge-work agents
24
20% measured
Computer use
16
estimated
Security & context learning
27
50% measured

Price

Input
$13.00/M
Output
$64.00/M
Cache read
$1.30/M
3:1 blend
$25.75/M
Agent-effective
$5.96/M

Agent-effective = 20 input tokens per output token, 85% cache hits. Why.

Agentic benchmarks

BenchmarkScoreNormalisedIndex weightHarness / source
SWE-bench Verified70.7%8%source
Terminal-Bench 2.0est.8%predicted 26
DeepSWEest.6%predicted 20
FrontierSWEest.5%predicted 3
CursorBenchest.4%predicted 24
GSO-Bench6.9%4%OpenHands
METR Time Horizons63.9%10%source
Vending-Bench 2est.6%predicted 9
PostTrainBenchest.4%predicted 16
APEX-Agentsest.8%predicted 20
GDPvalest.7%predicted 27
Remote Labor Indexest.5%predicted 1
DeepResearch Bench46.8%5%source
OSWorld 2.0est.6%predicted 2
OSWorldest.4%predicted 36
Cybench38.0%5%source
CL-benchest.5%predicted 16

Other benchmarks

BenchmarkCategoryScoreSource
Aider PolyglotCoding72.0%source
AlgoTuneCoding1.33×source
ARC-AGIReasoning35.7%source
ARC-AGI-2Reasoning8.6%source
CritPtReasoning0.3%source
DTBenchKnowledge81.6%source
Fiction.LiveBenchLong context61.1%source
GeoBenchMultimodal49.0%source
GPQA DiamondReasoning76.3%source
Humanity's Last ExamReasoning10.7%source
LMCAKnowledge37.4%source
MATH Level 5Math85.0%source
OTIS Mock AIMEMath64.4%source
SimpleBenchReasoning58.8%source
VPCTMultimodal38.0%source