pricing
Per-token, no rate-limit tiers.
USD per million tokens. Cached input is billed at the cached rate; every settled request has a usage receipt in the console.
| Model | Input | Cached input | Output |
|---|---|---|---|
kimi-k3 | $2.4 | $0.24 | $12 |
deepseek-v4.1-flash | $0.375 | $0.0075 | $1.5 |
muse-spark-open | $1.25 | $0.15 | $4.25 |
All models carry a 1,048,576-token context and 943,718 max output tokens. The machine-readable price list, including any long-context overrides, is the /models provider feed. Credits are purchased in the console; paid and promotional credits are tracked separately and new promotional credit does not expire.