MinT Service Level Agreement¶

Mind Lab Toolkit

Service Provider: MINDAI PTE. LTD.

Version: v2.0

Published: July [ ] 2026

Effective: [ ] 2026

PLEASE READ CAREFULLY. This Service Level Agreement ("SLA") sets out Mindai's service-availability and support commitments for MinT (Mind Lab Toolkit). It supplements, and is incorporated into, the MinT Terms of Service. Capitalised terms not defined herein have the meanings given in the Terms of Service.

ELIGIBILITY — This SLA applies ONLY to customers of the MinT Enterprise Edition ("Enterprise Customers") in respect of Covered Services for which Eligible Fees have been paid. It does NOT apply to the MinT Community Edition (free-tier users), to beta or preview features, or to services consumed under any free-trial or evaluation arrangement, all of which are provided on a best-efforts basis with no service-level commitment.

Article 1 — Scope and Eligibility¶

Mindai offers MinT in two editions:

Community Edition — provided free of charge for evaluation, learning, and non-production use. The Community Edition is provided AS IS, with no service-level commitment, no service credits, and no guaranteed support response time.
Enterprise Edition — provided to paying customers under an executed order or subscription. This SLA applies to the Enterprise Edition only.

This SLA covers only the Mindai Platform Layer of the Covered Services (as defined in Article 2). It does not extend to the IaaS Layer, to Customer-side systems, or to features expressly excluded from SLA coverage.

Article 2 — Definitions¶

"Covered Services" means the production-tier MinT services for which an Enterprise Customer has paid Eligible Fees, including: (a) the Training Platform's control plane API, dashboard, job-orchestration service, identity and access service, and metadata service operated by Mindai; and (b) the Model Gateway API layer operated by Mindai (including the API endpoints for the Macaron model series and third-party open-source models, and the rate-limiting and authentication layers) (collectively, the "Mindai Platform Layer").
"IaaS Layer" means the underlying compute, GPU, storage, network, and other infrastructure services provided by third-party cloud, GPU, or infrastructure providers (the "Subprocessors") on which MinT (comprising the Training Platform and the Model Gateway) runs.
"Eligible Fees" means the recurring service fees actually paid by Customer for the Covered Services in respect of the calendar month in which the SLA failure occurred, excluding one-off charges, professional-service fees, IaaS pass-through charges, taxes, and credits.
"Monthly Uptime Percentage" is calculated, for each calendar month, as: ((Total Minutes in the Month − Downtime − Excluded Time) ÷ (Total Minutes in the Month − Excluded Time)) × 100%.
"Downtime" means any period during which the Mindai Platform Layer of the Covered Services is unavailable to Customer for reasons within Mindai's reasonable control, as measured by Mindai's monitoring systems. Brief, transient errors not affecting the overall availability of the Covered Services are not counted.
"Excluded Time" has the meaning given in Article 7.
"Service Credit" means a credit, denominated in the currency of the original invoice, that may be applied against future invoices for the Covered Services in accordance with this SLA.
"Severity Level" or "P-Level" has the meaning given in Article 5.
"Standard Support Hours" means 09:00 to 18:00 Singapore Time (SGT, UTC+8), Monday through Friday, excluding public holidays in the Republic of Singapore.

Article 3 — Service Availability Commitment¶

Mindai will use commercially reasonable efforts to make the Mindai Platform Layer of the Covered Services available with a Monthly Uptime Percentage of at least 99.5% (the "Availability Target"), measured on a per-calendar-month basis.

Availability is measured by Mindai's monitoring systems at the control plane API and orchestration endpoints. Where Mindai's monitoring data conflicts with Customer's monitoring data, Mindai's monitoring data will be authoritative, save in cases of manifest error.

For the Model Gateway service, Mindai extends the availability target and commitments set out in this Article to the same standard as the Training Platform: the Model Gateway API layer accessed by enterprise-tier Customers is operated toward the same 99.5% monthly Availability Percentage target. Model Gateway availability measurement, Service Credit compensation, and Exclusions follow the corresponding provisions of this SLA. If a different availability target or independent measurement metric is later adopted for the Model Gateway, it will be separately agreed in a supplemental agreement or a subsequent version.

If Mindai fails to meet the Availability Target in any calendar month, Customer is eligible for a Service Credit in accordance with Article 4, subject to the claim process set out in Article 8 and the exclusions set out in Article 7.

Article 4 — Service Credits¶

Service Credits, where validly claimed and approved, are calculated as a percentage of the Eligible Fees for the affected Covered Services in the month in which the Availability Target was missed, in accordance with the table below:

Monthly Uptime Percentage	Service Credit (% of Eligible Fees)
Less than 99.5% but greater than or equal to 99.0%	5%
Less than 99.0% but greater than or equal to 95.0%	10%
Less than 95.0%	20% (cap)

The aggregate Service Credits payable to Customer in respect of any single calendar month shall not, under any circumstances, exceed twenty per cent (20%) of the Eligible Fees for that month, irrespective of the number of incidents or the cumulative Downtime.

Service Credits are non-refundable, non-transferable, and may be applied only against future invoices for the Covered Services. Service Credits are forfeited if (i) Customer's account is terminated for cause prior to credit application, or (ii) Customer fails to make timely payment of undisputed amounts due.

Article 5 — Support Response Times¶

Mindai will respond to support requests submitted by Enterprise Customers in accordance with the response targets set out below, measured during Standard Support Hours.

Severity	Definition	Initial Response Target
P1 (Critical)	Production environment of the Covered Services is wholly inaccessible or completely non-functional, with no commercially reasonable workaround.	Within 4 Standard Support Hours
P2 (High)	Major functionality of the Covered Services is materially impaired or significantly degraded; a workaround may exist but is impractical for sustained use.	Within 1 Standard Support Day
P3 (Medium)	Minor functionality is impaired; a reasonable workaround exists; production use is not materially affected.	Within 3 Standard Support Days
P4 (Low)	General questions, configuration assistance, documentation requests, or feature suggestions.	Within 5 Standard Support Days

"Initial Response" means an acknowledgement of the support request by a Mindai support engineer with an initial assessment and, where possible, an indicative path to resolution. It does not mean resolution of the underlying issue.

Support requests submitted outside Standard Support Hours are deemed received at the start of the next Standard Support Day for the purposes of measuring response times.

Severity classification is initially proposed by Customer and confirmed by Mindai in good faith. Where Mindai reasonably disagrees with the initial classification, Mindai will discuss with Customer and may re-classify the request, with reasons.

Extended support arrangements (including 24×7 coverage, accelerated P1 response, dedicated technical account management, named escalation paths, or on-site support) are available as a separately purchased add-on under a written extended-support order. Absent such an order, support is provided on the standard basis set out in this Article.

Article 6 — Maintenance Windows¶

From time to time, Mindai will perform scheduled maintenance to upgrade, patch, or improve the Covered Services. Mindai will use reasonable efforts to schedule such maintenance to minimise impact on Customer.

For Scheduled Maintenance, Mindai will provide Enterprise Customers with at least seven (7) days' prior notice via the dashboard or via email to the Customer's designated administrative contact, and will identify the expected start time, duration, and scope of the maintenance.

Mindai may also perform Emergency Maintenance where reasonably necessary to address a security vulnerability, integrity threat, or imminent service-impacting event. For Emergency Maintenance, Mindai will provide as much advance notice as is reasonably practicable in the circumstances.

Periods of Scheduled Maintenance and Emergency Maintenance are excluded from the calculation of Downtime and form part of Excluded Time.

Article 7 — Exclusions¶

"Excluded Time" means any period during which the Covered Services are unavailable, degraded, or interrupted as a result of any of the following, none of which counts as Downtime for purposes of calculating the Monthly Uptime Percentage:

(a) IaaS Layer events — outages, degradation, capacity constraints, latency, or other unavailability of any underlying compute, GPU, storage, network, or other infrastructure service operated by a Subprocessor. Mindai does not warrant the availability of, and is not responsible under this SLA for any unavailability caused by, the IaaS Layer;
(b) Force Majeure — any event beyond the reasonable control of Mindai, including without limitation acts of God, natural disasters, fire, flood, earthquake, epidemic or pandemic, war, civil unrest, terrorism, sabotage, strike, labour dispute, government order, embargo, sanction, restriction on cross-border data transfers, or denial-of-service attack of unusual magnitude;
(c) Scheduled Maintenance and Emergency Maintenance, as described in Article 6;
(d) Customer-caused issues — including any act, omission, configuration error, or misuse by Customer or Customer's personnel; defects in Customer-supplied training data, custom code, or container images; exceeding any quota, rate limit, fair-use threshold, or capacity reservation; or failure to comply with the technical requirements published by Mindai. For the avoidance of doubt, Mindai provides training-execution infrastructure and does not review, validate, or warrant the substantive quality, correctness, convergence, or fitness-for-purpose of Customer-authored training recipes, hyperparameters, reward functions, loss functions, or model architectures;
(e) Customer-side systems — issues arising from Customer's own equipment, software, networks, or third-party services not operated by Mindai;
(f) Suspension or termination of all or part of the Covered Services by Mindai under the Terms of Service or the Acceptable Use Policy, or in compliance with a lawful order of a competent authority;
(g) Beta, alpha, preview, experimental, or trial features expressly designated as such (collectively, "Pre-Release Features");
(h) Disruption arising from a Customer-initiated change to the configuration, integration, or environment of the Covered Services;
(i) The Community Edition and any other free-of-charge service tier; and
(j) Any other event expressly excluded from SLA coverage in the relevant order, statement of work, or product documentation.

Article 8 — Service Credit Claim Process¶

To receive a Service Credit, Customer must submit a written claim to contact@mindlab.ltd no later than thirty (30) calendar days after the end of the calendar month in which Mindai failed to meet the Availability Target. Claims submitted after this period will not be considered.

Each claim must include, at a minimum:

(a) Customer's account identifier and the affected Covered Services;
(b) the calendar month in respect of which the claim is made;
(c) Customer's calculation of the Monthly Uptime Percentage and the requested Service Credit;
(d) the dates, times, and duration of each Downtime incident relied upon, with available supporting evidence (e.g., error messages, monitoring screenshots, support ticket numbers); and
(e) confirmation that the incident is not, to Customer's knowledge, attributable to any of the categories of Excluded Time.

Mindai will validate each claim against its own monitoring records and respond with a decision within thirty (30) calendar days of receipt of a complete claim. Approved Service Credits will be applied against the next regular invoice issued to Customer for the Covered Services. No cash refunds will be made.

Article 9 — Sole and Exclusive Remedy¶

EXCEPT AS OTHERWISE EXPRESSLY PROVIDED IN THE TERMS OF SERVICE, SERVICE CREDITS AS SET OUT IN ARTICLE 4 ARE THE CUSTOMER'S SOLE AND EXCLUSIVE REMEDY, AND MINDAI'S SOLE AND EXCLUSIVE LIABILITY, FOR ANY FAILURE BY MINDAI TO MEET THE AVAILABILITY TARGET OR THE SUPPORT RESPONSE TARGETS SET OUT IN THIS SLA. Nothing in this Article 9 limits the parties' rights or remedies in respect of material breach of the Terms of Service to the extent such limitation would be unenforceable under applicable law.

Article 10 — Updates to This SLA¶

Mindai may update this SLA from time to time. Material updates that materially reduce Mindai's commitments to existing Enterprise Customers will take effect no earlier than thirty (30) days after notice to such Enterprise Customers, and will apply prospectively from the effective date. Customer's continued use of the Covered Services after the effective date of an update constitutes acceptance of the updated SLA.

Article 11 — Contact¶

All notices, support requests, and Service Credit claims under this SLA should be directed to:

Service Provider: MINDAI PTE. LTD., a private limited company incorporated in the Republic of Singapore with its registered office at 152 Beach Road, #11-05, Gateway East, Singapore 189721
Email (support requests, SLA claims, and general inquiries): contact@mindlab.ltd

Effective Date and Governing Versions¶

This SLA is effective from the date stated on the cover page. Mindai may publish localised translations of this SLA. In the event of any conflict between the English version and a localised translation, the English version shall prevail unless otherwise expressly required by applicable local law.

— End of Service Level Agreement —