SShogo Bench

Gemini 3.1 Flash Lite

Low-latency Gemini model for high-volume multimodal and agent workloads

Shogo Agent Index

Unranked

Not ranked yet: 1 agentic benchmark measured, and we need at least 2 plus an overall capability score. It will be ranked as results are published.

Agentic coding
31
estimated
Long-horizon autonomy
39
estimated
Knowledge-work agents
24
20% measured
Computer use
19
estimated
Security & context learning
32
estimated

Price

Input
$0.25/M
Output
$1.52/M
Cache read
—/M
3:1 blend
$0.57/M
Agent-effective
$0.31/M

Cheapest host for agents: Google at $0.060/M.

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

Agentic benchmarks

BenchmarkScoreNormalisedIndex weightHarness / source
SWE-bench Verifiedest.8%predicted 65
Terminal-Bench 2.0est.8%predicted 33
DeepSWEest.6%predicted 24
FrontierSWEest.5%predicted 4
CursorBenchest.4%predicted 26
GSO-Benchest.4%predicted 8
METR Time Horizonsest.10%predicted 64
Vending-Bench 2est.6%predicted 13
PostTrainBenchest.4%predicted 17
APEX-Agentsest.8%predicted 24
GDPvalest.7%predicted 30
Remote Labor Indexest.5%predicted 1
DeepResearch Bench37.3%5%source
OSWorld 2.0est.6%predicted 2
OSWorldest.4%predicted 44
Cybenchest.5%predicted 46
CL-benchest.5%predicted 17

Other benchmarks

BenchmarkCategoryScoreSource
CritPtReasoning1.1%source
DTBenchKnowledge76.8%source
FrontierMath Tiers 1-3Math27.7%source
GeoBenchMultimodal83.0%source
GPQA DiamondReasoning81.8%source
Humanity's Last ExamReasoning8.6%source
LMCAKnowledge35.0%source
OTIS Mock AIMEMath80.0%source
SciCodeCoding43.4%source
WebDev ArenaCoding1254source

API hosts

Live prices and uptime for every host serving Gemini 3.1 Flash Lite via OpenRouter.

HostInputOutputCache readAgent $/MUptime 24hQuant
Google$0.13$0.75$0.013$0.06070.7%—
Google AI Studio$0.13$0.75$0.013$0.060100.0%—
Google$0.25$1.50$0.025$0.13100.0%—
Google AI Studio$0.25$1.50$0.025$0.13100.0%—
Google$0.45$2.70$0.045$0.23100.0%—
Google AI Studio$0.45$2.70$0.045$0.23100.0%—