SShogo Bench

Kimi K2 0905

September 2025 Kimi K2 instruction update with a 256K context and improved agentic coding

Shogo Agent Index

Estimated#80 of 92
25.3range 16.2–34.3 · 12% of weight measured
Agentic coding
26
34% measured
Long-horizon autonomy
34
estimated
Knowledge-work agents
21
estimated
Computer use
13
estimated
Security & context learning
25
estimated

Price

Input
$1.06/M
Output
$3.17/M
Cache read
—/M
3:1 blend
$1.59/M
Agent-effective
$1.16/M

Cheapest host for agents: Novita at $0.69/M.

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

Agentic benchmarks

BenchmarkScoreNormalisedIndex weightHarness / source
SWE-bench Verifiedest.8%predicted 60
Terminal-Bench 2.027.8%8% Terminus 2
DeepSWEest.6%predicted 17
FrontierSWEest.5%predicted 2
CursorBenchest.4%predicted 22
GSO-Bench4.9%4%OpenHands
METR Time Horizonsest.10%predicted 59
Vending-Bench 2est.6%predicted 7
PostTrainBenchest.4%predicted 15
APEX-Agentsest.8%predicted 18
GDPvalest.7%predicted 24
Remote Labor Indexest.5%predicted 1
DeepResearch Benchest.5%predicted 43
OSWorld 2.0est.6%predicted 1
OSWorldest.4%predicted 31
Cybenchest.5%predicted 35
CL-benchest.5%predicted 15

Other benchmarks

BenchmarkCategoryScoreSource
Aider PolyglotCoding59.1%source
Fiction.LiveBenchLong context66.7%source
SimpleBenchReasoning26.3%source

API hosts

Live prices and uptime for every host serving Kimi K2 0905 via OpenRouter.

HostInputOutputCache readAgent $/MUptime 24hQuant
Novita$0.60$2.50—$0.69100.0%fp8