型号价格
当前启用的型号和美元价格
Pricing is informational. An individual price may be configured for each individual account.
Input-token caching and routed requestsBecause Model Gate can route requests across available upstream capacity, provider-side input-token or prompt caching may work, but it is not guaranteed for every request. When deterministic cache affinity is important, use a direct provider API account or a direct API key.
所有价格均以每百万代币美元为单位。该表是根据管理面板中启用的模型生成的。
| Model | Input | Output | Cache read | Cache write | Reasoning |
|---|---|---|---|---|---|
claude-opus-5 | $4 | $20 | $0.4 | $5 | $20 |
claude-sonnet-5 | $1.6 | $8 | $0.16 | $2 | $1.6 |
deepseek-v4-flash | $0.112 | $0.224 | $0.0022 | $0.112 | $0.224 |
deepseek-v4-pro | $0.348 | $0.696 | $0.0029 | $0.348 | $0.696 |
gpt-5.6-luna | $0.16 | $0.32 | $0.16 | $0.2 | $0.16 |
gpt-5.6-sol | $4 | $24 | $0.4 | $5 | $24 |
gpt-5.6-terra | $1.6 | $9.6 | $0.16 | $2 | $9.6 |
gpt-6-astra | $8 | $40 | $0.8 | $9.6 | $40 |
kimi-k2.7-code | $1.52 | $6.4 | $0.304 | $1.52 | $6.4 |
kimi-k3 | $2.4 | $12 | $0.24 | $2.4 | $12 |
mimo-v2.5 | $0.12 | $0.24 | $0.032 | $0.12 | $0.24 |
mimo-v2.5-pro | $0.36 | $0.72 | $0.032 | $0.36 | $0.72 |
minimax-m3 | $0.36 | $1.44 | $0.072 | $0.36 | $1.44 |
qwen3.7-max | $2 | $6 | $2 | $2 | $6 |
qwen3.7-plus | $0.96 | $3.84 | $0.96 | $0.96 | $3.84 |
claude-opus-4-6 | $4 | $20 | $0.4 | $5 | $20 |
claude-opus-4-7 | $4 | $20 | $0.4 | $5 | $20 |
claude-opus-4-8 | $4 | $20 | $0.4 | $5 | $20 |
claude-sonnet-4-6 | $2.4 | $12 | $0.24 | $3 | $12 |
deepseek-v4.1-flash | $0.12 | $0.48 | $0.024 | $0.12 | $0.48 |
gpt-5.4 | $2 | $12 | $0.2 | $3 | $12 |
gpt-5.5 | $4 | $24 | $0.4 | $4.8 | $24 |
qwen3.8-flash | $0.12 | $0.376 | $0.0128 | $0.16 | $0.376 |
qwen3.8-max | $1.6 | $4.8 | $0.2 | $2 | $4.8 |
普通表显示同步和本机 async=true 定价。如果您的定价模板的批量请求价格系数不是 1，页面在表格上方显示该系数。它仅适用于由 Model Gate 执行的 Claude/OpenAI 兼容批次项目。
您的帐户目录
经过认证的 型号 页面显示紧凑的行，其中包含每百万提供商代币的输入、输出和缓存读取价格（以美元为单位）。显示的账户费率包括适用于您账户的定价配置；官方提供商参考仅供参考，并非借记金额。只有当账户汇率较低时，官方价格才会被划掉。缺少的参考显示为未知而不是零价格。
按名称、型号 ID 或提供商搜索，按提供商/功能缩小范围，或按输出价格排序。 细节 打开模型的上下文限制、宣传的功能、模式和完整的五类速率表，包括缓存写入和推理。缺少的限制仍为“未发布”。明细表中的价格保留其存储的小数精度。个人账户还可以看到配置的代币显示系数；企业账户保留美元定价。