AI Model Catalog and Pricing

Groups and multipliers

claude-kiro（High cache）×0.15

Claude High cachehits

12 models available

claude-full×1.05

Claude full

12 models available

codex-full×0.27

Automatically detect and remove degraded accounts

5 models available

codex-value×0.07

Speed is not guaranteed; may be discontinued at any time

5 models available

codex-stable×0.12

Stable route

5 models available

DeepSeek×0.25

2 models available

grok×0.15

1 models available

Model rates

Model	Input (CNY / million)	Output (CNY / million)	Available groups
claude-fable-5 High cache	—	—	claude-fullclaude-kiro（High cache）
claude-fable-5-1 High cache	—	—	claude-fullclaude-kiro（High cache）
claude-haiku-4-5 High cache	—	—	claude-kiro（High cache）
claude-opus-4-5	—	—	claude-full
claude-opus-4-6 High cache	—	—	claude-fullclaude-kiro（High cache）
claude-opus-4-7 High cache	—	—	claude-fullclaude-kiro（High cache）
claude-opus-4-8 High cache	—	—	claude-fullclaude-kiro（High cache）
claude-opus-4.5 High cache	—	—	claude-kiro（High cache）
claude-opus-4.6 High cache	—	—	claude-kiro（High cache）
claude-opus-5 High cache	—	—	claude-fullclaude-kiro（High cache）
claude-opus-5-5 High cache	—	—	claude-fullclaude-kiro（High cache）
claude-sonnet-4-5	—	—	claude-full
claude-sonnet-4-6 High cache	—	—	claude-fullclaude-kiro（High cache）
claude-sonnet-5 High cache	—	—	claude-kiro（High cache）claude-full
claude-sonnet-5-5	—	—	claude-full
deepseek-v4-pro	—	—	DeepSeek
deepseek-v4.1-flash	—	—	DeepSeek
gpt-5.6-sol Multiple routesLong-context tiers	—	—	codex-Value routecodex-fullcodex-Stable route
gpt-5.6-terra Multiple routesLong-context tiers	—	—	codex-Value routecodex-fullcodex-Stable route
gpt-6-astra Multiple routesLong-context tiers	—	—	codex-Stable routecodex-Value routecodex-full
gpt-6-sol Multiple routesLong-context tiers	—	—	codex-Stable routecodex-Value routecodex-full
gpt-6.1-sol Multiple routesLong-context tiers	—	—	codex-Value routecodex-fullcodex-Stable route
grok-4.7	—	—	grok

How are charges calculated?
Charge =（Input token × input rate ＋ Output token × output rate ＋ cached tokens × applicable cache rates）÷ 1,000,000 × 500,000 × group multiplier。
For deepseek-v4-pro ：Weekday China Standard Time 9:00–12:00、14:00–18:00 forpeak hours，other hours are off-peak，the two periods have different rates（see the table above「peak/off-peak」tag）。
Long context（beyond 272K token）SomeModelis charged at a higher tier。Cache reads and writes use their respective rates.。

23 models. Clear, published rates.

Groups and multipliers

Model rates

SelectedModel?Create a key to get started