generated: '2026-08-26' method: searched source: >- https://docs.hyperstack.cloud/docs/release-notes, https://docs.hyperstack.cloud/docs/support, https://status.hyperstack.cloud/, https://docs.hyperstack.cloud/docs/api-reference/introduction - cross-derived against openapi/nexgen-cloud-hyperstack-openapi.json versioning: style: uri-path current: v1 base_url: https://infrahub-api.nexgencloud.com/v1 policy_published: false note: >- No written versioning policy exists. The API has carried /v1 since launch; the spec and SDKs move on an independent 1.54.x line that is not surfaced in the URI or in any response header. A consumer cannot tell which build of the contract they are talking to. deprecation_policy: published: false sunset_header: false deprecation_header: false rfc8594: false advance_notice: not stated note: >- Hyperstack has no written deprecation policy and emits no RFC 8594 Sunset or Deprecation headers. Deprecation happens - the 2026-08-25 release note withdrew AI Studio fine-tuning, model deployment, evaluations, datasets and data synthesis from the API on the day it was announced, along with four named base models - but it is announced only as prose in the release notes, with no machine-readable signal, no notice period and no migration window. This is the single largest operational gap for an autonomous consumer: an agent has no runtime way to learn that an operation it depends on is going away. observed_deprecations: - date: '2026-08-25' scope: AI Studio withdrawn: - fine-tuning - LoRA adapter training and import - model deployment - model aliases - model evaluations - training metrics and logs - datasets - data synthesis - 'retired models: openai/gpt-oss-120b, mistralai/Mistral-Small-24B-Instruct-2501, meta-llama/Llama-3.3-70B-Instruct, meta-llama/Llama-3.1-8B-Instruct' notice: none - effective on announcement source: https://docs.hyperstack.cloud/docs/release-notes deprecated_operations: count: 0 note: >- Not one operation in either published OpenAPI carries `deprecated: true`. Verified across all 209 Hyperstack operations and all 33 AI Studio operations. This is not a clean bill of health - billing-history operations for the withdrawn AI Studio features (Get_fine-tuning_billing_history, Get_data_synthesis_billing_history, Get_model_evaluation_billing_history, Get_resource_fine-tuning_billing_history, Get_resource_model_evaluation_billing_history) are still declared as live in the contract the day after the features they bill for were removed. status_page: published: true url: https://status.hyperstack.cloud/ http_status: 200 incidents_feed: not published maintenance_notices: true note: Linked from the docs support page as "Service Status". sla: published: false note: >- No public SLA or uptime commitment. Availability commitments for Secure Private Cloud are contracted, not published. health_endpoints: - operationId: Get_alive path: /billing/alive scope: billing API liveness - operationId: healthzEndpoint path: /object-storage/healthz scope: object storage API liveness - url: http://localhost:8080/health scope: self-hosted API MCP Server liveness (documented response includes service version) support: channels: - {type: ticket, url: 'https://nexgencloudsupport.zendesk.com/hc/en-gb', note: 'Support Center (Zendesk). Returns 403 to a plain crawler UA; live for a browser.'} - {type: email, value: support@hyperstack.cloud} - {type: docs, url: 'https://docs.hyperstack.cloud/docs/support'} security_advisories: published: true example: id: CVE-2026-31431 title: 'Security Advisory: Copy Fail' url: https://docs.hyperstack.cloud/docs/support/cve-2026-31431 summary: Linux kernel privilege-escalation flaw and Hyperstack VM mitigation steps. note: >- The provider publishes named CVE advisories with mitigation steps in its own docs. It does NOT publish a security.txt or a vulnerability-disclosure policy - see security/nexgen-cloud-vulnerability-disclosure.yml. changelog: url: https://docs.hyperstack.cloud/docs/release-notes detail: changelog/nexgen-cloud-changelog.yml