Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.apiyi.com/llms.txt

Use this file to discover all available pages before exploring further.

​
Quick Answer

Yes—available based on your needs. Objectively speaking, no provider in today’s AI industry can promise a 99.99% SLA (not even OpenAI or Claude themselves). APIYI does not make promises it can’t keep; instead, it offers guarantees that actually work: abnormal-billing compensation for high-volume image customers, credit reissuance for losses caused by our issues, and SLA terms with compensation caps that can be set in enterprise contracts.

​
An Honest Premise

Why not simply promise 99.99%?AI evolves rapidly—upstream models frequently update, rate-limit, and adjust their policies. Even OpenAI and Claude themselves cannot achieve a 99.99% SLA. This is an objective fact across the entire industry.Rather than offering a number we can’t deliver, we prefer to provide clear, executable guarantee mechanisms that ensure customers don’t suffer losses when issues arise.

​
The SLA Guarantees We Provide

Abnormal-Billing Compensation for High-Volume Image Customers

For high-volume customers of Nano Banana Pro / 2 and GPT-Image-2 (monthly spend of $10,000+), we verify and compensate clearly abnormal billing (such as a request that timed out yet was still billed).

Credit Reissuance for Our Issues

For losses clearly caused by us (e.g., backend server anomalies), we export logs uniformly and reissue credits so users don’t suffer losses for nothing.

SLA in Enterprise Contracts

Enterprise customers can stipulate SLA terms and compensation caps in the contract—a clear safeguard for both parties.

Transparent Real-Time Status

Service status and incident announcements are published publicly on “Live Updates”, so customers can stay informed about upstream and platform conditions.

​
1. Abnormal-Billing Compensation for High-Volume Image Customers

For high-volume customers of image models such as Nano Banana Pro / 2 and GPT-Image-2 (monthly spend of $10,000 or more can participate), we provide compensation for clearly abnormal billing. For example: an image request that abnormally times out (e.g., beyond 300 seconds) yet is still billed as a generated image—for such clearly abnormal billing, we verify the logs and compensate, so high-volume customers don’t pay for nothing.
This is just an example scenario, not a generation-time promise.We cannot promise “an image within 300 seconds, or we compensate”—AI image generation is subject to upstream fluctuations, so we can’t make a hard commitment on generation speed. This guarantee targets losses from clearly abnormal billing, not generation speed. Contact support to confirm the exact scope.

​
2. Credit Reissuance for Losses Caused by Our Issues

If a loss is clearly caused by us, we proactively compensate. Typical scenarios:
  • A backend server anomaly causing customer credit loss
  • An image model abnormal timeout that still returned data and was billed
Our commitment: for situations where a clear issue on our side affects customers, we will export logs uniformly, verify the affected scope, and reissue credits to affected customers, ensuring users don’t suffer losses for nothing.
Attribution principle: This reissuance applies to issues clearly attributable to us (e.g., platform backend anomalies). Factors outside our control—such as upstream official content moderation blocks or upstream rate limiting—are handled per their respective explanations (for example, some upstream moderation blocks are not billed in the first place).

​
3. SLA Stipulated in Enterprise Contracts

For enterprise customers, the specific SLA terms can be explicitly stipulated in the contract, including:
  • Coverage scope and availability metrics
  • Incident response and resolution timeframes
  • Terms such as a compensation cap
This way, SLA guarantees and responsibility boundaries are clearly defined at signing—a clear safeguard for both the customer and APIYI.

​
Frequently Asked Questions

Upstream AI models frequently update, rate-limit, and adjust. Even OpenAI and Claude themselves cannot achieve 99.99%—this is an objective fact. We don’t make promises we can’t keep; instead, we protect customer interests with clear, executable compensation mechanisms.
This guarantee is for high-volume customers of image models such as Nano Banana Pro / 2 and GPT-Image-2; those with monthly spend of $10,000 or more can participate, covering clearly abnormal billing (for example, a request that times out yet is still billed as a generated image). This is not a “guaranteed image within 300s or we pay” promise—it’s compensation for abnormal-billing losses. Contact support to enable and set it up.
For losses clearly caused by us (e.g., backend server anomalies, image model abnormal timeouts that returned data), we will export logs uniformly, verify the affected scope, and reissue credits so users don’t suffer losses for nothing.
Yes. Enterprise customers can stipulate SLA terms and compensation caps in the contract, so coverage and responsibility boundaries are clearly defined at signing—a safeguard for both parties.

Live Updates

Service status and incident announcements

How do enterprise customers recharge?

Corporate transfer and contract signing

Refund Policy

Original-path refund rules and application process

Difference between enterprise and individual users?

Enterprise accounts, multi-token, and internal sharing

​
Contact Us

Enterprise WeChat Support

For SLA activation, contract term agreements, credit reissuance, and more, contact us: