SShogo Bench

Qwen3.5 35B-A3B

Qwen vision-language model for visual reasoning, documents, and agent tasks

Shogo Agent Index

Unranked

Not ranked yet: 0 agentic benchmarks measured, and we need at least 2 plus an overall capability score. It will be ranked as results are published.

Agentic coding
29
estimated
Long-horizon autonomy
38
estimated
Knowledge-work agents
24
estimated
Computer use
17
estimated
Security & context learning
30
estimated

Price

Input
$0.25/M
Output
$2.00/M
Cache read
—/M
3:1 blend
$0.69/M
Agent-effective
$0.33/M

Cheapest host for agents: Darkbloom at $0.070/M.

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

Agentic benchmarks

BenchmarkScoreNormalisedIndex weightHarness / source
SWE-bench Verifiedest.8%predicted 64
Terminal-Bench 2.0est.8%predicted 30
DeepSWEest.6%predicted 22
FrontierSWEest.5%predicted 3
CursorBenchest.4%predicted 25
GSO-Benchest.4%predicted 7
METR Time Horizonsest.10%predicted 62
Vending-Bench 2est.6%predicted 11
PostTrainBenchest.4%predicted 16
APEX-Agentsest.8%predicted 22
GDPvalest.7%predicted 28
Remote Labor Indexest.5%predicted 1
DeepResearch Benchest.5%predicted 44
OSWorld 2.0est.6%predicted 2
OSWorldest.4%predicted 40
Cybenchest.5%predicted 44
CL-benchest.5%predicted 16

Other benchmarks

BenchmarkCategoryScoreSource
CritPtReasoning0.6%source
DTBenchKnowledge80.0%source
GPQA DiamondReasoning83.5%source
LMCAKnowledge29.5%source
OTIS Mock AIMEMath70.0%source
SciCodeCoding29.3%source

API hosts

Live prices and uptime for every host serving Qwen3.5 35B-A3B via OpenRouter.

HostInputOutputCache readAgent $/MUptime 24hQuant
Darkbloom$0.065$0.75$0.033$0.07098.9%fp4
DeepInfra$0.11$0.80$0.040$0.0900.0%fp8
DeepInfra$0.14$1.00$0.050$0.1198.6%fp8
Parasail$0.15$1.00$0.050$0.1199.9%fp8
Venice$0.15$1.00$0.050$0.1199.9%—
Alibaba$0.16$1.30—$0.22100.0%—
SiliconFlow$0.24$1.80$0.15$0.2473.5%fp8
AtlasCloud$0.23$1.80$0.23$0.3093.3%fp8