SShogo Bench

MiniMax-M2

MiniMax model for chat, coding, office work, and agentic tasks

Shogo Agent Index

Estimated#83 of 92
23.9range 16–31.9 · 14% of weight measured
Agentic coding
26
23% measured
Long-horizon autonomy
31
30% measured
Knowledge-work agents
20
estimated
Computer use
12
estimated
Security & context learning
23
estimated

Price

Input
$0.35/M
Output
$1.41/M
Cache read
—/M
3:1 blend
$0.61/M
Agent-effective
$0.40/M

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

Agentic benchmarks

BenchmarkScoreNormalisedIndex weightHarness / source
SWE-bench Verifiedest.8%predicted 58
Terminal-Bench 2.030.0%8%Terminus 2
DeepSWEest.6%predicted 15
FrontierSWEest.5%predicted 2
CursorBenchest.4%predicted 21
GSO-Benchest.4%predicted 4
METR Time Horizonsest.10%predicted 57
Vending-Bench 2$1616%source
PostTrainBenchest.4%predicted 14
APEX-Agentsest.8%predicted 16
GDPvalest.7%predicted 23
Remote Labor Indexest.5%predicted 1
DeepResearch Benchest.5%predicted 43
OSWorld 2.0est.6%predicted 1
OSWorldest.4%predicted 28
Cybenchest.5%predicted 32
CL-benchest.5%predicted 15

Other benchmarks

BenchmarkCategoryScoreSource
SimpleBenchReasoning25.0%source
WebDev ArenaCoding1298source