SShogo Bench

Qwen3.6 Max Preview

Flagship Qwen model for complex reasoning, coding, and agentic workflows

Shogo Agent Index

Estimated#55 of 92
40.5range 29.2–51.9 · 14% of weight measured
Agentic coding
42
23% measured
Long-horizon autonomy
56
30% measured
Knowledge-work agents
31
estimated
Computer use
27
estimated
Security & context learning
40
estimated

Price

Input
$2.15/M
Output
$12.86/M
Cache read
$0.20/M
3:1 blend
$4.83/M
Agent-effective
$1.08/M

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

Agentic benchmarks

BenchmarkScoreNormalisedIndex weightHarness / source
SWE-bench Verified76.7%8%source
Terminal-Bench 2.0est.8%predicted 53
DeepSWEest.6%predicted 35
FrontierSWEest.5%predicted 8
CursorBenchest.4%predicted 31
GSO-Benchest.4%predicted 16
METR Time Horizonsest.10%predicted 70
Vending-Bench 2$4,2546%source
PostTrainBenchest.4%predicted 21
APEX-Agentsest.8%predicted 34
GDPvalest.7%predicted 38
Remote Labor Indexest.5%predicted 2
DeepResearch Benchest.5%predicted 46
OSWorld 2.0est.6%predicted 5
OSWorldest.4%predicted 61
Cybenchest.5%predicted 61
CL-benchest.5%predicted 19

Other benchmarks

BenchmarkCategoryScoreSource
DTBenchKnowledge87.2%source
GPQA DiamondReasoning87.4%source
LMCAKnowledge42.5%source
OTIS Mock AIMEMath91.1%source
SimpleBenchReasoning63.0%source
SimpleQA VerifiedKnowledge52.0%source
WebDev ArenaCoding1479source