August 2026

August 28

IONOS CLOUD supports Managed Kubernetes version 1.35. For more information about the schedule timeline, see Release Schedule.

August 28

Qwen3.8 27B is a compact, dense multimodal model from Alibaba Cloud's Qwen team, now available in the IONOS CLOUD AI Model Hub. As the most capable generation in the Qwen open-model family to date, it pairs 27B parameters with an integrated vision encoder and a 262K-token context window. The model accepts text and image inputs, supports tool calling, and offers configurable reasoning depth through the reasoning_effort parameter (none, low, medium, xhigh), with medium as the default. This makes it well-suited for complex, multi-step reasoning and agentic workloads. Video input is available in Beta.

August 18

IONOS CLOUD has launched a new data center, de/fra/1, in the Frankfurt region. It is directly networked to the existing de/fra and de/fra/2 locations, ensuring high availability, failover redundancy, and operational security. The data center runs on Intel Sierra Forest and AMD EPYC Turin processors.

The new data center appears in the Data Center Designer (DCD) location selector as Germany / Frankfurt am Main (de/fra/1) and is available via the Cloud API as de/fra/1.

August 17

IONOS CLOUD Tracing Service is now available. It provides a centralized and scalable solution for collecting, storing, and analyzing distributed traces from your applications, services, and infrastructure.

Create a tracing pipeline to receive an authenticated ingestion endpoint for your traces. Each pipeline supports OTLP/HTTP and OTLP/gRPC ingestion protocols, a trace ingestion key, and a Grafana endpoint for trace exploration. The service stores traces for a 30-day retention period. Billing is based on trace ingestion volume (GB) and trace storage (GB).

August 13

Information on the security advisory for CrackArmor is now available in the documentation portal. Disclosed on March 19, 2026, this vulnerability affects Managed Kubernetes node images and allows an attacker with code execution inside a container to escape the container boundary and gain access to the underlying host operating system.

IONOS CLOUD has applied updated Managed Kubernetes node images across all regions to mitigate the vulnerability.

August 13

IONOS CLOUD Logging Service and Monitoring Service now support the OpenTelemetry Protocol (OTLP) for ingesting telemetry data.

August 12

IONOS CLOUD supports Confidential VM, a hardware-enforced security technology that encrypts virtual machine (VM) memory and CPU state at runtime using AMD SEV-SNP (Secure Encrypted Virtualization, Secure Nested Paging), ensuring that data in use remains protected from the cloud provider, hypervisor, and host administrators. Workloads run inside a Trusted Execution Environment (TEE), where sensitive data and code are cryptographically isolated within the AMD CPU and inaccessible to unauthorized users even at the infrastructure level.

Organizations handling regulated data, including financial services, healthcare, and public sector workloads, can use IONOS CLOUD Confidential VMs to meet zero-trust security requirements and enforce data sovereignty. Confidential VMs can be attested using the IONOS CLOUD reference toolchain. For more information, see SNPGuard.

August 6

To improve the IONOS CLOUD partner experience and help ensure compliance with Marketplace guidelines, the IONOS CLOUD Marketplace now includes a product approval process.

Before publishing a product to the Marketplace, partners with an active Partner Network profile submit a product approval request through the Partner Portal. The Account Management team reviews each request and publishes approved products to the Marketplace, providing a rejection reason if a request is not approved so partners can address the issue and resubmit. For more information, contact your dedicated Partner Account Manager.

August 31

A new guide lists the AI Model Hub error codes with the cause and solution for each, grouped into client, model, capacity, and server errors. It also covers the OpenAI-format 529 overload response, how to handle errors using the openai Python library, and what to include when reporting a persistent error to support.

August 27

New metrics documentation is available for PostgreSQL and In-Memory DB v2 clusters, covering infrastructure metrics (shared across database types) and database-specific health metrics.

August 17

A new tutorial shows how to export n8n workflow and node execution traces to the IONOS CLOUD Tracing Service using n8n's built-in OpenTelemetry support, with no code changes. It covers creating a tracing pipeline, configuring n8n through the settings UI or environment variables, running a workflow, and exploring the resulting workflow.execute and node.execute spans in Grafana with TraceQL.

August 17

A new tutorial shows how to instrument a Python AI agent that calls IONOS CLOUD AI Model Hub and export GenAI traces to the Tracing Service. It covers creating a tracing pipeline, instrumenting the agent with the OpenAI OpenTelemetry instrumentation, and exploring gen_ai.* span attributes such as the model, token usage, latency, and prompts in Grafana with TraceQL.

August 12

A new end-to-end tutorial for deploying a Confidential VM with SNPGuard is now available. The tutorial covers attestation server deployment, converting a base image to a confidential format with LUKS2 encryption, registering the expected measurement with the attestation server, and creating the Confidential VM using the IONOS CLOUD API. It demonstrates a full workload integrity lifecycle, from image preparation through runtime attestation, using the SNPGuard reference toolchain.

August 5

A new migration guide is available for IONOS CLOUD Compute Engine. The guide covers how to move between vCPU Server, Dedicated Core Server, and Cube server types, including the Storage Drag & Drop and Snapshot methods, using the IONOS CLOUD Backup Service, and the BIOS/UEFI firmware compatibility rules required to avoid start failures.

August 12

Llama 3.1 8B is deprecated and will be removed from the IONOS CLOUD AI Model Hub on October 1, 2026. We recommend migrating to Qwen3.5 9B, its designated successor. After removal, requests using this model ID will receive an HTTP 400 Bad Request response with an "Invalid model name" error, instead of being served by the model.

August 6

Llama 3.1 405B is deprecated and will be removed from the IONOS CLOUD AI Model Hub on September 15, 2026. We recommend migrating to Qwen3.5 397B A17B, its designated successor. After removal, requests using this model ID will receive an HTTP 400 Bad Request response with an "Invalid model name" error, instead of being served by the model.