Moonshot AI Kimi Code k2.7

2 appearances · meta-score 67.4/100 · last seen 2026-07-10 · scored under rubric v15

Pooled composite across 2 primary runs: 60.2 ± 11.6 (95% CI) provisional: n < 3, treat as unsettled

$43.48Total cost
151,171,278Tokens
88.0%Cache hit
2,458Requests
0.00%Failure rate
0.93Tools/response
3,477,119Tokens per $

Agent role scores

Per-agent-role scores for Moonshot AI Kimi Code k2.7
Agent role	Score / 10
Sprint-Plan Agent	7.55
Pseudocode Agent	7.48
Sprint-Review Agent	7.32
Scout Agent	7.22
Spec Agent	6.92
Develop Agent	6.72
Code-Review Agent	6.62
Code Agent	6.53
Orchestrator	5.90
Test Agent	5.84
Platform Agent	5.32
Build Agent	5.08

Run history (primary model)

Runs where Moonshot AI Kimi Code k2.7 was the primary model, oldest first
Run	SoW	Date	Composite
Kimi Code k2.7	circles	Jul 2026	61.1
Kimi Code k2.7	circles	Jul 2026	59.3

Every run on Favur Evals is scored by the same deterministic engine, on the same Statements of Work. The benchmark is self-funded — no vendor sponsorship, credits, or grants.