SShogo Bench

GPT-6.1 Terra

Fast, efficient model for visual reasoning, precise computer use, and clear, natural writing.

Shogo Agent Index

Estimated#8 of 92
69.0range 60.5–77.5 · 14% of weight measured
Agentic coding
79
estimated
Long-horizon autonomy
74
30% measured
Knowledge-work agents
56
32% measured
Computer use
63
estimated
Security & context learning
62
estimated

Price

Input
$1.00/M
Output
$5.00/M
Cache read
$0.050/M
3:1 blend
$2.00/M
Agent-effective
$0.42/M

Cheapest host for agents: OpenAI at $0.21/M.

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

Agentic benchmarks

BenchmarkScoreNormalisedIndex weightHarness / source
SWE-bench Verifiedest.8%predicted 89
Terminal-Bench 2.0est.8%predicted 96
DeepSWEest.6%predicted 80
FrontierSWEest.5%predicted 58
CursorBenchest.4%predicted 54
GSO-Benchest.4%predicted 75
METR Time Horizonsest.10%predicted 86
Vending-Bench 2$9,1076%source
PostTrainBenchest.4%predicted 39
APEX-Agents68.3%8%source
GDPvalest.7%predicted 70
Remote Labor Indexest.5%predicted 19
DeepResearch Benchest.5%predicted 53
OSWorld 2.0est.6%predicted 42
OSWorldest.4%predicted 95
Cybenchest.5%predicted 94
CL-benchest.5%predicted 30

Other benchmarks

BenchmarkCategoryScoreSource
CritPtReasoning28.0%source
FrontierMath Tiers 1-3Math90.9%source
FrontierMath Tier 4Math92.7%source
GPQA DiamondReasoning94.3%source
MirrorCodeCoding76.7%source
OTIS Mock AIMEMath100.0%source
SciCodeCoding55.1%source
SimpleBenchReasoning72.2%source
SimpleQA VerifiedKnowledge57.3%source
WeirdML v3Coding28.1%codex_cli

API hosts

Live prices and uptime for every host serving GPT-6.1 Terra via OpenRouter.

HostInputOutputCache readAgent $/MUptime 24hQuant
OpenAI$0.50$2.50$0.025$0.21100.0%—
Azure$1.00$5.00$0.050$0.42100.0%—
OpenAI$1.00$5.00$0.050$0.42100.0%—
OpenAI$2.00$10.00$0.10$0.84100.0%—