Model economics / detail
kimi-k2p6S
Dispatch tier: Inexpensive (complexity class this model is routed for)
54 runs, 54 successes, zero kills, timeouts, or losses — a flawless sheet at real volume, done cheap and fast. Others hauled more freight; nobody else hauled it without dropping a single crate.
The measured record
What the ledger shows
Fleet-ledger aggregates for this model only — run counts, outcomes, spend, and token appetite. Missing telemetry reads “Not observed”, never zero.
Agent runs
54
Success rate
100%
Metered spend
$14.64
31 of 54 runs metered
Cost / metered run
$0.47
Tokens in
59.3M
across 31 token-measured runs
Tokens out
435.1k
Tokens in / run
1.9M
Tokens out / run
14.0k
Outcomes
Status breakdown
Every run ends in exactly one status. The bar is the whole record, to scale.
No non-success statuses on record — every run succeeded.
Run duration
How long the runs take
Wall-clock distribution across measured runs, from the fastest exit to the longest grind.
Min
60s
p25
3m
Median
5m
p75
8m
p90
14m
p99
18m
Max
20m
Mean
7m
Duration measured on 54 of 54 runs.
Commentary
Idiosyncrasies
What stands out in this model's numbers — shape, appetite, and failure habits.
- 54 runs, 54 successes, zero failures, kills, timeouts, or losses — the only flawless sheet at real volume.
- Never finished under a minute (min 60s): no fast-fail runs at all, which may be the whole secret.
- Inexpensive-tier economics to match: ~$0.47 per metered run, though only 31 of 54 runs carried token/cost telemetry.
Commentary
Lessons learned
Practical routing and operations takeaways, grounded in the same record.
- A perfect record at 54 runs and workhorse prices earns more volume — the open question is whether the streak survives deepseek-scale traffic.
- Whatever keeps it from dying in the first minute, the aliases and siblings that do fast-fail could stand to learn it.