Seedance 2.5实测：用ModelGo每条10秒视频省$0.31

四组不同的生产需求、两个API平台，以及实际费用记录。本组测试中，ModelGo合计费用为$8.038059，OpenRouter为$9.283320；在实际交付规格一致的条件下，ModelGo费用低13.41%。

Seedance 2.5是什么？

Seedance 2.5支持文字驱动和参考图驱动的视频创作，适合包含细致动作要求的制作需求。可用于短叙事、受控形态变化和产品装配演示。

测试方法与实际费用

本次测试包含四组独立需求：两组纯文字生成、两组参考图生成；在双平台可访问时，使用相同Prompt、参考文件和请求设置。ModelGo费用来自逐条匹配的已结算账单，OpenRouter费用来自已完成响应中的用量费用；表格展示模型输出费用，保留六位小数，不包含单独生成参考图的费用。

费用口径：本文比较模型调用费用，不是为账户充值所需的实际支出；不包含充值或支付手续费，包括另一篇价格对比文章计入的5.5%充值费。本文费用差额及比例仅按调用费用计算。

四个真实生产案例

为准确复现真实调用，本节保留当时提交的英文Prompt原文；中文说明用于解释生产目标。

货运自行车晨间配送故事

应用行业：城市出行/品牌广告。

十秒内展示三箱货运自行车转弯、停在面包店门口、交付一箱的完整故事；服装、箱数和车轮运动须连续，避免危险交通或品牌露出。

实际提交的英文Prompt

Create a 10-second commercial micro-story in a quiet early-morning neighborhood. A fictional teal electric cargo bicycle carries three secured kraft delivery boxes; an adult rider wears a plain mustard jacket and helmet. Seconds 0–4: a low tracking shot follows the bike around one gentle corner, with realistic wheel rotation and stable box placement. Seconds 4–7: the rider slows and stops beside a small bakery entrance. Seconds 7–10: a restrained wider shot shows the rider handing one box to an adult bakery owner, with the other two boxes still secured. Use warm morning light and natural city sound if audio is generated. No real brands, traffic hazards, text overlays or changing clothing. Deliver one coherent ten-second clip. Avoid speed ramps, unrelated cutaways and abrupt changes in camera direction; preserve the specified subject throughout.

ModelGo费用低13.41%。

苔藓微景观环保品牌短片

应用行业：环保品牌/叙事。

镜头沿苔藓微景观低位前移，跟随一滴水落入溪流，最后仰拍揭示桌上的陶瓷托盘；保持植物形态和溪流走向，不夸大环保效益。

实际提交的英文Prompt

Create a 10-second macro-to-landscape illusion for an environmental brand film. The scene is a miniature moss garden with a shallow stream, tiny fern fronds and one smooth dark pebble. Begin at moss-tip level with a slow low dolly, making the plants feel like a vast forest without changing their actual scale. Seconds 3–7: follow a single water droplet as it slides from a fern tip into the stream, creating small concentric ripples. Seconds 7–10: tilt up gently to reveal the complete garden inside a shallow ceramic tray on a studio table. Keep all plant shapes and stream geography consistent. Soft daylight, restrained natural ambience, no fantasy creatures, text or exaggerated environmental claims. Deliver one coherent ten-second clip. Avoid speed ramps, unrelated cutaways and abrupt changes in camera direction; preserve the specified subject throughout.

ModelGo费用低13.41%。

平板家具装配演示

应用行业：家具/产品说明。

保留胡桃木桌面与两条弯曲胶合板桌腿，先分离展示，再缓慢滑入对应槽位，最后落地小幅转镜；不增加部件或将概念动画当作安装验证。

这张原创参考图用于ModelGo；对OpenRouter可访问的模型，对照请求也使用同一文件。

实际提交的英文Prompt

Animate the supplied original flat-pack side table into a clear 10-second assembly concept film. Preserve the walnut top, two curved plywood legs, slot geometry and pale background. Seconds 0–3: show the tabletop and two legs separated but aligned above their final positions. Seconds 3–7: the two legs slide into their matching underside slots in a slow readable sequence, with no bending or shape changes. Seconds 7–10: the completed table settles onto the ground and the camera makes a small three-quarter reveal. Use soft studio shadows and realistic wood grain. No hands, extra screws, additional components, text or claims that this is a verified installation instruction. Prioritize mechanical coherence and part identity. Deliver one coherent ten-second clip. Avoid speed ramps, unrelated cutaways and abrupt changes in camera direction; preserve the specified subject throughout.

ModelGo费用低13.41%。

纸鹤变礼盒品牌短片

应用行业：包装/品牌短片。

从靛蓝外侧、金色内侧的参考纸鹤开始，缓慢展翼、展开并重新折成开口礼盒，最后俯视定格；只能使用同一张纸，不增添纸张或消失碎片。

这张原创参考图用于ModelGo；对OpenRouter可访问的模型，对照请求也使用同一文件。

实际提交的英文Prompt

Create a 10-second square brand film beginning from the supplied original folded-paper crane on an ivory desk. Preserve the paper’s indigo outer face, gold inner face and crisp fold pattern. Seconds 0–4: a slow macro move follows a gentle opening of the wings. Seconds 4–8: the same sheet unfolds in a controlled origami motion and refolds into a small open-top geometric gift box, keeping its two paper colors consistent and its material visibly paper. Seconds 8–10: the camera eases back to a stable overhead hero frame with the open box centered and clean surrounding space. Use quiet paper rustle if audio is generated. No added sheets, disappearing scraps, logos, hands or text. Deliver one coherent ten-second clip. Avoid speed ramps, unrelated cutaways and abrupt changes in camera direction; preserve the specified subject throughout.

ModelGo费用低13.41%。

如何计算费用差额

对于实际交付规格一致的输出，费用降低比例的计算方式为(OpenRouter 实际费用 − ModelGo 已结算费用) / OpenRouter 实际费用 × 100。判断两次输出是否可比时，应检查实际返回的尺寸、时长和音轨。这四组需求的平均单次费用差为$0.311315；如果按同样工作负载重复100或1,000次，费用差约为$31.131529或$311.315294。这只是按本次实测推算的用量示例，不是新增实际账单。

开始使用ModelGo API

先安装requests并设置环境变量MODELGO_API_KEY。以下代码使用首个真实案例的端点、模型ID和参数，已在代码中直接写入实际Prompt。此示例提交异步视频任务，后续轮询及下载方式见代码下方。

import os, requests

prompt = 'Create a 10-second commercial micro-story in a quiet early-morning neighborhood. A fictional teal electric cargo bicycle carries three secured kraft delivery boxes; an adult rider wears a plain mustard jacket and helmet. Seconds 0–4: a low tracking shot follows the bike around one gentle corner, with realistic wheel rotation and stable box placement. Seconds 4–7: the rider slows and stops beside a small bakery entrance. Seconds 7–10: a restrained wider shot shows the rider handing one box to an adult bakery owner, with the other two boxes still secured. Use warm morning light and natural city sound if audio is generated. No real brands, traffic hazards, text overlays or changing clothing. Deliver one coherent ten-second clip. Avoid speed ramps, unrelated cutaways and abrupt changes in camera direction; preserve the specified subject throughout.'
headers = {"Authorization": "Bearer " + os.environ["MODELGO_API_KEY"]}
payload = {'model': 'doubao-seedance-2.5',
 'prompt': prompt,
 'resolution': '720p',
 'duration': 10,
 'aspect_ratio': '16:9',
 'generate_audio': True}
response = requests.post('https://api.modelgo.ai/v1/video/generations', headers=headers, json=payload, timeout=420)
response.raise_for_status()
result = response.json()
print(result)  # Save the returned task ID for polling.

提交后保存响应中的id / task_id，通过GET https://api.modelgo.ai/v1/videos/{task_id}轮询至成功或失败状态；成功后使用响应中的metadata.url下载视频。

原生content端点的首帧生成请求使用content数组，其中包含文字部分，以及带有role: "first_frame"的image_url部分。本篇参考图案例应同时省略ratio和aspect_ratio，输出画面比例由参考图决定。

ModelGo与OpenRouter的模型ID和媒体请求格式存在差异。可以复用创作需求，接入时请根据各平台文档调整端点、参考图结构、任务轮询和下载处理。

ModelGo API文档

在ModelGo使用Seedance 2.5

使用本页介绍的Seedance 2.5模型，请前往ModelGo的Models页面。

前往ModelGo使用Seedance 2.5

参考资料

以下资料用于核对模型能力、接口与用量记录；案例需求由本次测试独立编写。

• 模型官方文档或系列展示：用于了解能力与创意方向；相关系列资料不代表其他变体的规格保证。
• OpenRouter API参考：用于核对请求约定及平台差异。
• OpenRouter用量与费用记录：用于理解用量信息；本文费用来自实际测试记录。
• OpenRouter模型目录API：在新接入前核对模型ID与目录属性。
• OpenRouter视频生成指南：用于了解媒体请求结构和工作流。