AI 模型价格 - 当前 AIsa LLM 与媒体模型费率

本页列出网关中全部 108 个模型的当前公开 AIsa Model Gateway 价格，来自实时定价数据，刷新时间为 2026 年 8 月 5 日。模型能力、上下文窗口和 endpoint 映射请查看支持的模型目录。所有 token 价格均以 USD / 100 万 tokens 标价。按请求计费的模型会按生成资产或调用计费。Workspace 级别的价格规则可能会改变 Usage Logs 中显示的最终金额。

计费公式

Total Cost = (input_tokens / 1,000,000 * input_price) + (output_tokens / 1,000,000 * output_price) 只有当上游 route 返回对应计费 bucket 时，cache read 和 cache write 价格才会适用。

OpenAI

Model ID	输入 / 1M	输出 / 1M	缓存读取 / 1M	缓存写入 / 1M
`gpt-4.1-mini`	$0.4211	$1.6842	$0.1053	-
`gpt-4o`	$2.6316	$10.5263	$1.3158	-
`gpt-4o-mini`	$0.1579	$0.6316	$0.0789	-
`gpt-5-chat-latest`	$1.3158	$10.5263	$0.1316	-
`gpt-5-nano`	$0.0526	$0.4211	$0.0053	-
`gpt-5-pro`	$15.7895	$126.3158	-	-
`gpt-5-search-api`	$1.3158	$10.5263	$0.1316	-
`gpt-5.1`	$1.3158	$10.5263	$0.1316	-
`gpt-5.1-chat-latest`	$1.3158	$10.5263	$0.1316	-
`gpt-5.1-codex`	$1.3158	$10.5263	$0.1316	-
`gpt-5.1-codex-max`	$1.3158	$10.5263	$0.1316	-
`gpt-5.2`	$1.8421	$14.7368	$0.1842	-
`gpt-5.2-chat-latest`	$1.8421	$14.7368	$0.1842	-
`gpt-5.2-codex`	$1.8421	$14.7368	$0.1842	-
`gpt-5.2-pro`	$22.1053	$176.8421	-	-
`gpt-5.3-chat-latest`	$1.8421	$14.7368	$0.1842	-
`gpt-5.3-codex`	$1.8421	$14.7368	$0.1842	-
`gpt-5.4`	$2.6316	$15.7895	$0.2632	-
`gpt-5.4-mini`	$0.7895	$4.7368	$0.0789	-
`gpt-5.4-nano`	$0.2105	$1.3158	$0.0211	-
`gpt-5.4-pro`	$31.5789	$189.4737	-	-
`gpt-5.5`	$5.2632	$40.0000	$0.5263	-
`gpt-5.5-pro`	$31.5789	$189.4737	-	-
`gpt-5.6-luna`	$0.2105	$1.2632	$0.0211	$0.2500
`gpt-5.6-sol`	$5.2632	$31.5789	$0.5263	$6.2500
`gpt-5.6-terra`	$2.1053	$12.6316	$0.2105	$2.5000
`gpt-image-2`	$8.4211	$31.5789	-	$2.0000

Model ID	价格	Endpoint(s)
`gpt-image-2`	$8.4211 输入 / $31.5789 输出，每 1M Token（缓存写入 $2.0000/M）；$0.03 / 次请求; $0.05 / 张图 (1024x1024)	`/v1/images/generations`, `/v1/images/edits`

Anthropic

Model ID	输入 / 1M	输出 / 1M	缓存读取 / 1M	缓存写入 / 1M
`claude-fable-5`	$10.5263	$52.6316	$1.0526	$12.5000
`claude-haiku-4-5-20251001`	$1.0526	$5.2632	$0.1053	$2.0000
`claude-opus-4-1-20250805`	$15.7895	$78.9474	$1.5789	$30.0000
`claude-opus-4-5-20251101`	$5.2632	$26.3158	$0.5263	$10.0000
`claude-opus-4-6`	$5.2632	$26.3158	$0.5263	$10.0000
`claude-opus-4-7`	$5.2632	$26.3158	$0.5263	$10.0000
`claude-opus-4-8`	$5.2632	$26.3158	$0.5263	$10.0000
`claude-opus-5`	$5.2632	$26.3158	$0.5263	$10.0000
`claude-sonnet-4-5-20250929`	$3.1579	$15.7895	$0.3158	$6.0000
`claude-sonnet-4-6`	$3.1579	$15.7895	$0.3158	$6.0000
`claude-sonnet-4-6-thinking`	$3.1579	$15.7895	$0.3158	$6.0000
`claude-sonnet-5`	$2.1053	$10.5263	$0.2105	$2.5000

Google Gemini

Model ID	输入 / 1M	输出 / 1M	缓存读取 / 1M	缓存写入 / 1M
`gemini-3-pro-preview`	$2.1053	$12.6316	-	-
`gemini-3.5-flash`	$1.5789	$9.4737	$0.1579	-

xAI

Model ID	输入 / 1M	输出 / 1M	缓存读取 / 1M	缓存写入 / 1M
`grok-4.20-0309-non-reasoning`	$1.3158	$2.6316	-	-
`grok-4.20-0309-reasoning`	$1.3158	$2.6316	-	-
`grok-4.3`	$1.3158	$2.6316	-	-
`grok-4.5`	$2.1053	$6.3158	-	$0.5000
`grok-build-0.1`	$1.0526	$2.1053	-	-

DeepSeek

Model ID	输入 / 1M	输出 / 1M	缓存读取 / 1M	缓存写入 / 1M
`deepseek-r1`	$0.4229	$1.6903	$0.4229	-
`deepseek-v3`	$0.2115	$0.8452	$0.2115	-
`deepseek-v3.1`	$0.4229	$1.2681	$0.4229	-
`deepseek-v3.2`	$0.2115	$0.3176	$0.2115	-
`deepseek-v4-flash`	$0.1032	$0.2063	$0.0021	-
`deepseek-v4-flash-0731`	$0.1380	$0.2750	-	$0.0280
`deepseek-v4-pro`	$0.3205	$0.6411	$0.0027	-

Alibaba

Model ID	输入 / 1M	输出 / 1M	缓存读取 / 1M	缓存写入 / 1M
`qwen-flash`	$0.0154	$0.1512	$0.0035	-
`qwen-mt-flash`	$0.0720	$0.2205	$0.0720	-
`qwen-mt-lite`	$0.0840	$0.2520	$0.0840	-
`qwen-plus-2025-12-01`	$0.2800	$0.8400	$0.2800	-
`qwen3-coder-480b-a35b-instruct`	$1.0500	$5.2500	$1.0500	-
`qwen3-coder-flash`	$0.2100	$1.0500	$0.2100	-
`qwen3-coder-plus`	$0.7000	$3.5000	$0.7000	-
`qwen3-max`	$0.7200	$3.6000	$0.7200	-
`qwen3-vl-flash`	$0.0350	$0.2800	$0.0350	-
`qwen3-vl-flash-2025-10-15`	$0.0350	$0.2800	$0.0350	-
`qwen3-vl-plus`	$0.1400	$1.1200	$0.1400	-
`qwen3.6-plus`	$0.2760	$1.6510	$0.2760	-
`qwen3.6-plus-2026-04-02`	$0.2760	$1.6510	$0.2760	-
`qwen3.7-flash`	$0.1732	$0.6930	$0.0347	$0.2166
`qwen3.7-flash-2026-07-15`	$0.1732	$0.6930	$0.0347	$0.2166
`qwen3.7-max`	$1.1550	$3.4657	$0.1155	$1.4441
`qwen3.7-max-2026-06-08`	$1.1550	$3.4657	$0.2310	-
`qwen3.7-plus`	$0.1932	$0.7707	$0.0392	-
`qwen3.8-max`	$1.7325	$5.1986	$0.2163	$2.1662

Model ID	价格	Endpoint(s)
`wan2.7-i2v`	$0.060208 / 输出秒 (720p); $0.100347 / 输出秒 (1080p)	`/v1/video/generations`
`wan2.7-image`	$0.03 / 次请求	`/v1/images/generations`, `/v1/images/edits`
`wan2.7-image-pro`	$0.075 / 次请求	`/v1/images/generations`, `/v1/images/edits`
`wan2.7-r2v`	$0.060208 / 输出秒 (720p); $0.100347 / 输出秒 (1080p)	`/v1/video/generations`
`wan2.7-t2v`	$0.060208 / 输出秒 (720p); $0.100347 / 输出秒 (1080p)	`/v1/video/generations`

Moonshot

Model ID	输入 / 1M	输出 / 1M	缓存读取 / 1M	缓存写入 / 1M
`kimi-k2-thinking`	$0.4229	$1.6903	$0.4229	-
`kimi-k2.5`	$0.4229	$2.2186	$0.0847	-
`kimi-k2.6`	$0.6587	$2.7360	$0.6587	-
`kimi-k2.7-code`	$0.6257	$2.5992	$0.1252	-
`kimi-k3`	$3.1579	$15.7895	$0.3158	-

MiniMax

Model ID	输入 / 1M	输出 / 1M	缓存读取 / 1M	缓存写入 / 1M
`MiniMax-M2.5`	$0.2211	$0.8842	$0.2211	-
`MiniMax-M3`	$0.2211	$0.8842	$0.0526	-

Zhipu GLM

Model ID	输入 / 1M	输出 / 1M	缓存读取 / 1M	缓存写入 / 1M
`glm-5`	$0.4222	$1.9011	$0.4222	-
`glm-5.1`	$0.5775	$2.3107	$0.1155	-
`glm-5.2`	$0.7700	$2.6957	$0.1925	-
`z-ai/glm-5v-turbo`	$1.2000	$4.0000	$0.2400	-

ByteDance

Model ID	输入 / 1M	输出 / 1M	缓存读取 / 1M	缓存写入 / 1M
`seed-1-6-250915`	$0.2368	$0.9474	$0.2368	-
`seed-1-6-flash-250715`	$0.0711	$0.2842	$0.0711	-
`seed-1-8-251228`	$0.2368	$1.8947	$0.2368	-
`seed-2-0-lite-260228`	$0.2632	$2.1053	$0.2632	-
`seed-2-0-mini-260215`	$0.1053	$0.4211	$0.1053	-
`seed-2-0-pro-260328`	$0.5263	$3.1579	$0.5263	-

Model ID	价格	Endpoint(s)
`dreamina-seedance-2-0-260128`	$4.3 / 1M Token (480p, 720p, 4k); $4.7 / 1M Token (1080p)	`/v1/video/generations`
`dreamina-seedance-2-0-fast-260128`	$3.3 / 1M Token (480p, 720p)	`/v1/video/generations`
`seedream-4-5-251128`	$0.036 / 次请求	`/v1/chat/completions`
`seedream-5-0-260128`	$0.035 / 次请求	`/v1/chat/completions`, `/v1/messages`, `/v1/responses`, `/v1/images/generations`

Xiaomi

Model ID	输入 / 1M	输出 / 1M	缓存读取 / 1M	缓存写入 / 1M
`mimo-v2.5`	$0.1505	$0.3011	$0.0031	-
`mimo-v2.5-pro`	$0.5158	$0.9021	$0.0038	-

StepFun

Model ID	输入 / 1M	输出 / 1M	缓存读取 / 1M	缓存写入 / 1M
`step-3.5-flash`	$0.1053	$0.3158	$0.0211	-
`step-3.5-flash-2603`	$0.1053	$0.3158	$0.0211	-
`step-3.7-flash`	$0.2105	$1.2105	$0.0421	-

HappyHorse

Model ID	价格	Endpoint(s)
`happyhorse-1.1-i2v`	$0.098 / 输出秒（720p）；$0.126 / 输出秒（更高档；上游数据把两档都标为 720p）	`/v1/video/generations`
`happyhorse-1.1-r2v`	$0.098 / 输出秒 (720p); $0.126 / 输出秒 (1080p)	`/v1/video/generations`
`happyhorse-1.1-t2v`	$0.098 / 输出秒 (720p); $0.126 / 输出秒 (1080p)	`/v1/video/generations`

其他

Model ID	输入 / 1M	输出 / 1M	缓存读取 / 1M	缓存写入 / 1M
`text-embedding-3-small`	$0.0211	-	-	-
`text-embedding-3-large`	$0.1368	-	-	-

重要说明

生产环境变更前，请以 aisa.one/models 上的实时可用性和价格为准。
Token 计费模型按输入和输出用量计费。缓存读取和缓存写入价格仅在上游路由上报这些计费项时才适用。
图像模型按请求或按生成的图片计费。
视频模型有两种计费口径：Wan 和 HappyHorse 路由按输出秒数计费，不同分辨率单价不同；即梦 Seedance 路由按 1M tokens 计费，不同分辨率单价不同。
Embedding 模型仅按输入 token 计费。
gpt-image-2 同时公布 token 价格和按请求 / 按图档位，实际适用哪一档取决于你调用的路由。
每次调用的最终计费金额可在 AIsa Usage Logs 页面查看。

AIsa 文档

产品导览

教程

价格

AIsa 控制台

学习

对比

Agent 发现

更新日志

AI 模型价格 - 当前 AIsa LLM 与媒体模型费率

计费公式

OpenAI

Anthropic

Google Gemini

xAI

DeepSeek

Alibaba

Moonshot

MiniMax

Zhipu GLM

ByteDance

Xiaomi

StepFun

HappyHorse

其他

重要说明

​计费公式

​OpenAI

​Anthropic

​Google Gemini

​xAI

​DeepSeek

​Alibaba

​Moonshot

​MiniMax

​Zhipu GLM

​ByteDance

​Xiaomi

​StepFun

​HappyHorse

​其他

​重要说明

计费公式

OpenAI

Anthropic

Google Gemini

xAI

DeepSeek

Alibaba

Moonshot

MiniMax

Zhipu GLM

ByteDance

Xiaomi

StepFun

HappyHorse

其他

重要说明