Comparison
eroq vs AI/ML API
AI/ML API sells breadth: hundreds of mainstream models, one endpoint, familiar per-token pricing. eroq sells depth on one vertical: the character product, from uncensored roleplay to image, video, voice and hosting. Which one fits depends entirely on what you are building.
免费赠送 40 积分 · 无需绑卡 · 所有账号均支持私密模式
Where an aggregator shines
If your product is general-purpose — assistants, summarization, coding tools — an aggregator's catalogue breadth and model-swapping are genuinely useful, and AI/ML API's developer experience is competent. For light roleplay their mainstream catalogue also works, with the caveat every mainstream model carries: the filter.
Where the vertical platform wins
Character products fail on the details aggregators cannot own:
- →Refusal behavior: mainstream models refuse dark or intense scenes probabilistically — mid-scene, differently after every update. eroq's engines render within a written policy, with deterministic content_blocked errors.
- →Roleplay tuning: persona hold, repetition damping and register control measured on production companion traffic — not a system prompt taped over a general model.
- →The full stack: uncensored images, async video, voice both ways, CDN storage, and a Private mode that keeps nothing.
- →Flat credits: heavy chat users are the norm in this vertical; per-token pricing makes them a liability, flat pricing makes them your best customers.
Side by side
The short version:
| eroq | AI/ML API | |
|---|---|---|
| Built for | Character & creative products | General-purpose model access |
| Catalogue | A focused roster across chat, image, video and voice | Hundreds of mainstream models |
| Content posture | Uncensored within a written AUP | Mainstream filters apply |
| Pricing | Flat credits, auto-refunds | Per-token tiers |
| Media & storage | Images · video · voice · CDN | Not the focus |
| Best for | Companion & character apps with revenue | General AI features |
Public positioning as of August 2026 — verify current details on aimlapi.com. General product, aggregator; character product, vertical platform. Mixing both across one codebase is normal.
一次调用,立见效果
curl https://eroq.ai/v1/chat/completions \
-H "Authorization: Bearer $EROQ_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "eroq-rp-plus",
"context": "Mira: sardonic starship mechanic, dry humor, hates small talk. Scene: engine bay, mid-shift.",
"messages": [
{
"role": "user",
"content": "The reactor is making that noise again."
}
]
}'Base URL https://eroq.ai/v1 · 完整参考见 文档
| 模型 | 类型 | 价格 |
|---|---|---|
| eroq-rp-plus | chat | 3 积分 / 次回复 |
| eroq-rp-mini | chat | 1 积分 / 次回复 |
| claude-fable-5-2 | chat | 20 积分 / 次回复 |
| eroq-uncensored | image | 10 积分 / 张 |
| eroq-krea2 | image | 10 积分 / 张 |
| eroq-one | image | 10 积分 / 张,每多一张图片 +1 |
| eroq-anime | image | 10 积分 / 张 |
| eroq-motion-one | video | 29 积分 / 秒 |
| eroq-motion-pro | video | 30 积分 / 秒 |
| seedance-2-5 | video | 98 积分 / 秒 |
| veo-3-1 | video | 169 积分 / 秒 |
| kling-3-pro | video | 71 积分 / 秒 |
| seedance-2-0 | video | 64 积分 / 秒 |
| hailuo-3 | video | 55 积分 / 秒 |
| runway-gen-4-5 | video | 51 积分 / 秒 |
| veo-3-1-fast | video | 43 积分 / 秒 |
| kling-3 | video | 54 积分 / 秒 |
| seedance-2-0-fast | video | 39 积分 / 秒 |
| hailuo-3-max | video | 34 积分 / 秒 |
| grok-imagine | video | 60 积分 / 秒 |
| seedance-2-0-mini | video | 15 积分 / 秒 |
| veo-3-1-lite | video | 22 积分 / 秒 |
| wan-3 | video | 43 积分 / 秒 |
| eroq-flash-image | image | 10 积分 / 张 |
| eroq-flux-image | image | 12 积分 / 张 |
| gpt-image-2-5-sunburst | image | 31 积分 / 张 |
| gpt-image-2-5-flare | image | 31 积分 / 张 |
| nano-banana-2-1 | image | 31 积分 / 张 |
| seedream-5-0-pro | image | 36 积分 / 张 |
| nano-banana-pro | image | 124 积分 / 张 |
| flux-3-image | image | 32 积分 / 张 |
| grok-imagine-2-0 | image | 32 积分 / 张 |
完整价格表见 价格页面。
开发者常见问题
Can AI/ML API handle uncensored roleplay? +
Its catalogue is mainstream models, and mainstream models carry content filters — intense scenes refuse probabilistically. eroq's engines are selected and tuned so fiction renders, governed by a written policy instead.
Is eroq OpenAI-compatible like the aggregators? +
Yes — chat/completions, images/generations and audio/speech follow the OpenAI request convention, so code written for an aggregator ports by changing the base URL, key and model id.
Which has more models? +
AI/ML API, by two orders of magnitude — that is the aggregator's job. eroq deliberately runs a small set of engines it can tune and guarantee for one vertical.
What about hosting the media my product generates? +
That is the difference in kind: eroq ships CDN storage (2 credits per 10MB, once) wired directly into generation via store: true, under the same written policy as the engines. Aggregators leave hosting to you.
