SShogo Bench

Qwen3 VL 30B A3B Instruct

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
—
estimated
Long-horizon autonomy
—
estimated
Knowledge-work agents
—
estimated
Computer use
—
estimated
Security & context learning
—
estimated

Price

Input
$0.15/M
Output
$0.55/M
Cache read
$0.030/M
3:1 blend
$0.25/M
Agent-effective
$0.070/M

Cheapest host for agents: DeepInfra at $0.17/M.

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

Agentic benchmarks

BenchmarkScoreNormalisedIndex weightHarness / source
SWE-bench Verifiedest.8%—
Terminal-Bench 2.0est.8%—
DeepSWEest.6%—
FrontierSWEest.5%—
CursorBenchest.4%—
GSO-Benchest.4%—
METR Time Horizonsest.10%—
Vending-Bench 2est.6%—
PostTrainBenchest.4%—
APEX-Agentsest.8%—
GDPvalest.7%—
Remote Labor Indexest.5%—
DeepResearch Benchest.5%—
OSWorld 2.0est.6%—
OSWorldest.4%—
Cybenchest.5%—
CL-benchest.5%—

API hosts

Live prices and uptime for every host serving Qwen3 VL 30B A3B Instruct via OpenRouter.

HostInputOutputCache readAgent $/MUptime 24hQuant
DeepInfra$0.15$0.60—$0.17100.0%fp8
Novita$0.20$0.70—$0.2295.6%bf16
SiliconFlow$0.29$1.00—$0.3295.2%fp8