Per-token price is the wrong way to compare AI models for agentic work.
Kimi K3 lists at $15 per million output tokens. On the Artificial Analysis Index it spends 130M output tokens against a 63M median, so on the same suite at the same list rate it bills like $31/M.
The per-token price is that it hides a 2x spread in what a task actually costs.
Kimi K3 lists at $15 per million output tokens. On the Artificial Analysis Index it spends 130M output tokens against a 63M median, so on the same suite at the same list rate it bills like $31/M.
The per-token price is that it hides a 2x spread in what a task actually costs.
4
4
25
2,890
Verbosity is trainable. In one post-training pass, DeepSeek’s V4-Flash-0731 cut output tokens 12% and gained 10 index points. The 28M tokens it dropped weren’t buying the score.
Whether K3’s verbosity costs you depends on the comparator. On a median-verbosity workload at list K3 $1,950, Opus 5 $1,575. On measured cost per index task, K3 is cheapest of the three: $0.86 vs $2.03 for Opus 5 and $3.15 for Fable 5.
Full breakdown in our State of Open Source AI report. stateofopensource.ai
Whether K3’s verbosity costs you depends on the comparator. On a median-verbosity workload at list K3 $1,950, Opus 5 $1,575. On measured cost per index task, K3 is cheapest of the three: $0.86 vs $2.03 for Opus 5 and $3.15 for Fable 5.
Full breakdown in our State of Open Source AI report. stateofopensource.ai
Oct 14, 2026 · 8:29 PM UTC
0
1
5
826