Terms and Conditions

Effective October 4, 2026. By purchasing a token package, activating an API key, or using our API, you agree to the following terms.

Payments are accepted via DuitNow. By completing a purchase, you agree to these terms and the usage policy as enforced by our gateway.

Read the full Usage Policy for details on how our gateway screens requests and enforces acceptable use.

1. About the service

Our gateway provides paid access to multiple artificial intelligence models through a compatible API at api.mireld.my/v1. Some service functions depend on third-party models, networks, and infrastructure.

Supported endpoints: /v1/chat/completions, /v1/completions, /v1/responses, /v1/messages, /v1/messages/count_tokens, and /v1/models. Availability, features, capacity, speed, and output quality may differ between models and may change due to technical conditions or third-party policies.

All token packages receive identical model access. Package sizes differ only in the number of credits provided.

2. API keys and security

You are responsible for keeping your API key secure and for all activity performed through it. Keys must not be shared publicly or committed to public repositories.

3. Credits, packages, and pricing

Credit packages are sold in fixed sizes (50M, 100M, 200M, 300M, 500M, and 1B credits). Every request deducts credits according to the applicable model rate and actual token usage. Usage is billed on exact token counts; failed requests due to upstream errors are not charged.

Each key’s validity period is 30 days, starting from the first successful billable request. Unused credits expire when the validity period ends.

The administrator reserves the right to change model rates, credit prices, and package prices at any time. Changes are announced at the time they take effect through our official channels and apply only to purchases and usage after the change. Changes never reduce credits already recorded in your balance.

Requests are subject to the rate limit and concurrency limit set on your key at issuance (default 40 requests per minute, 10 simultaneous requests).

4. Refunds

Refund requests are accepted for any reason. Where a refund is granted, the amount is calculated from the remaining unused credits and the remaining validity of the key.

The administrator retains sole discretion in deciding whether a refund is granted and the refund amount. Usage already processed is not refundable. Refunds are issued through the original payment method where possible, and payment gateway fees that are not returned may be deducted.

5. Model availability and disruptions

Specific models may experience disruptions, slow responses, timeouts, interrupted responses, or discontinuation at any time. We use reasonable efforts to maintain the service but do not guarantee uninterrupted availability of every model.

When a request is routed through an alternative path or model, the output, speed, and capabilities of the response are not guaranteed to be identical. Check request status before retrying to avoid duplicate requests.

We may add, restrict, or discontinue models at any time. Changes are announced at the time they take effect.

6. AI model output

Model output may be inaccurate, incomplete, biased, outdated, or unsafe to execute directly. You must review output before relying on it, particularly for medical, legal, financial, or security decisions.

We do not guarantee that model output is correct, unique, free of third-party rights, or suitable for a particular purpose.

7. Prohibited use

The service must not be used for unlawful conduct, fraud, phishing, malware creation, credential theft, system attacks, bypassing authentication or billing, infringement of third-party rights, or other abuse that harms our service or third parties.

Attempts to circumvent, bypass, disable, or probe the limits of safety guardrails and usage policies to obtain restricted or prohibited output are not allowed.

8. Violations and suspensions

Requests are screened before they reach any model. High-confidence violations are blocked and recorded against your key with escalating suspensions: 30 minutes for the first violation, 1 hour for the second, and 24 hours for the third and subsequent violations within the escalation ladder. Violation records decay after 7 days of clean usage.

Suspensions apply per API key. Requests made while a key is suspended are rejected.

9. Limitation of liability

To the extent permitted by law, we are not liable for indirect losses, lost profits, lost opportunities, or data loss resulting from use of the service or decisions based on model output.

Our direct liability is limited to the amount paid by you during the 3 months preceding the directly related claim, unless applicable law prohibits that limitation.

10. Service changes and termination

We may add, replace, restrict, or discontinue service features, including for scheduled or emergency maintenance. Where an account is closed without a violation on your part, unused credits remain eligible for refund consideration under section 4.

These terms may be updated to reflect changes to the service, security, or law. Material changes are announced through our official channels.

11. Governing law and dispute resolution

These terms are governed by the laws of Malaysia. Disputes should first be raised through the official contact channel, without limiting any dispute-resolution rights available to you under applicable law.

12. Contact

Questions, complaints, and refund requests are handled through our official Telegram administrator bot.