SShogo Bench

GPT-5 Mini

Small GPT-5 for responsive agents, coding help, and everyday automation

Shogo Agent Index

Estimated#69 of 92
30.4range 21.4–39.3 · 22% of weight measured
Agentic coding
32
46% measured
Long-horizon autonomy
36
30% measured
Knowledge-work agents
26
estimated
Computer use
21
estimated
Security & context learning
33
estimated

Price

Input
$0.25/M
Output
$2.00/M
Cache read
$0.025/M
3:1 blend
$0.69/M
Agent-effective
$0.15/M

Cheapest host for agents: OpenAI at $0.080/M.

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

Agentic benchmarks

BenchmarkScoreNormalisedIndex weightHarness / source
SWE-bench Verified64.7%8%source
Terminal-Bench 2.034.8%8%spoox-o-m
DeepSWEest.6%predicted 26
FrontierSWEest.5%predicted 5
CursorBenchest.4%predicted 27
GSO-Benchest.4%predicted 9
METR Time Horizonsest.10%predicted 65
Vending-Bench 2$-316%source
PostTrainBenchest.4%predicted 18
APEX-Agentsest.8%predicted 26
GDPvalest.7%predicted 31
Remote Labor Indexest.5%predicted 1
DeepResearch Benchest.5%predicted 45
OSWorld 2.0est.6%predicted 3
OSWorldest.4%predicted 47
Cybenchest.5%predicted 49
CL-benchest.5%predicted 17

Other benchmarks

BenchmarkCategoryScoreSource
AlgoTuneCoding1.38×source
ARC-AGIReasoning54.3%source
ARC-AGI-2Reasoning4.4%source
Chess PuzzlesReasoning24.0%source
CritPtReasoning0.0%source
DTBenchKnowledge80.5%source
Fiction.LiveBenchLong context69.4%source
FrontierMath Tiers 1-3Math46.7%source
FrontierMath Tier 4Math12.2%source
GPQA DiamondReasoning75.0%source
Humanity's Last ExamReasoning19.4%source
LMCAKnowledge34.2%source
MATH Level 5Math97.8%source
OTIS Mock AIMEMath86.7%source
ProofBenchMath9.0%source
SciCodeCoding39.2%source
SimpleQA VerifiedKnowledge21.6%source
VPCTMultimodal40.2%source

API hosts

Live prices and uptime for every host serving GPT-5 Mini via OpenRouter.

HostInputOutputCache readAgent $/MUptime 24hQuant
OpenAI$0.13$1.00$0.013$0.080100.0%—
OpenAI$0.25$2.00$0.025$0.15100.0%—
Azure$0.25$2.00$0.030$0.16100.0%—