> For the complete documentation index, see [llms.txt](https://docs.stg.ai.neevcloud.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.stg.ai.neevcloud.com/release-notes/june-2026-release.md). # June 2026 Release **Release Type:** Major Feature Release\ **Release Version:** v0.5.0 *** ## Overview The June 2026 release is our most ambitious to date, introducing **Agentic Studio** — NeevCloud's managed environment for building, running, and hosting AI agents at scale. This release brings: * Managed Sandboxes for secure and isolated agent code execution * An Agents hosting platform with a curated OSS template catalogue * Bring-Your-Own-Model support for dedicated GPU inference * Expanded developer access via CLI, SDK, and MCP — all authenticated through Personal Access Tokens * Full TCP and UDP networking for CPU Virtual Machines Together, these capabilities make NeevCloud the platform of choice for teams building production-grade agentic AI applications. *** ## What's New ### Agentic Studio — Sandboxes *(Beta)* Sandboxes give AI agents a secure, managed environment to execute code, read and write files, call external APIs, and maintain state across multiple turns — without any infrastructure setup on your end. Each sandbox is fully isolated, provisioned with configurable CPU, memory, and disk, and accessible via dedicated APIs. You can create, manage, and monitor sandboxes through the platform, with usage billed on-demand. This release ships with a complete sandbox management API (create, list, get, and delete sandboxes with your preferred resource configuration), file and execution APIs for reading and writing files and running commands directly within a live sandbox, a sandbox template catalogue of pre-configured environments to get started quickly, a real-time health dashboard to monitor sandbox status at a glance, and per-organisation and per-project quota enforcement to keep resource usage predictable. Full API reference, quickstart guides, and code examples are available in the [Sandbox documentation](https://docs.ai.neevcloud.com/agentic-studio/). **What's coming:** Go SDK, expanded template options, and configurable egress controls. **Note:** Sandboxes are in Beta. The feature is fully functional and available to all users, but is under active development. See the Understanding Beta section below for details. ### Agentic Studio — Agents *(Beta)* NeevCloud introduces Agents — a hosted platform for deploying and running AI agents directly on NeevCloud infrastructure. Agents are tightly integrated with NeevCloud's inference layer and Sandboxes, giving you a complete agentic stack in one place. This release ships with an OSS Agent template catalogue (pre-built templates for popular agents including OpenClaw and Codex CLI, with more templates being added in upcoming releases), agent lifecycle management to deploy, update, and tear down agent instances via API, secure agent access through scoped access control and gateway routing that keeps each agent within the correct project and organisation boundaries, and usage-based billing with agent compute tracked and billed per-resource, visible in your billing history. **What's coming:** An interactive agent console and UI for managing and interacting with agents directly from the platform. **Note:** Agents are in Beta. The feature is fully functional and available to all users, but is under active development. See the Understanding Beta section below for details. ### AI Inference — Bring Your Own Model You can now deploy your own model weights as a dedicated GPU inference endpoint on NeevCloud — not just models from the NeevCloud catalogue. Any HuggingFace-compatible model can be brought in and deployed through the standard [Dedicated Model Deployment](https://docs.ai.neevcloud.com/ai-inference/) workflow, giving teams with proprietary or fine-tuned models full control over what runs in production. ### Developer Access — CLI, SDK & MCP NeevCloud now offers a full suite of developer access tools, all authenticated through **Personal Access Tokens (PAT)**. Whether you prefer working from the terminal, integrating via API, or connecting through an MCP-compatible client, you can manage and interact with the platform programmatically. * **CLI (`neev-cli`)** — manage GPU workloads, AI inference, sandboxes, storage, and more directly from your terminal. * **SDK** — access the full range of platform APIs programmatically from your application or automation pipeline. Authentication is handled via API key, and the SDK is accessible regardless of your stack. * **MCP** — connect to NeevCloud services through any MCP-compatible client, enabling seamless integration with AI development workflows and tooling. All three are secured through PAT-based authentication. You can generate and manage your Personal Access Tokens from the platform under API Key Management. ### VM Service — Full TCP & UDP Networking Support CPU Virtual Machines on NeevCloud now support both **TCP and UDP** protocols, giving you complete control over how your VMs communicate within the platform and with the outside world. * **TCP support** — reliable, connection-oriented traffic for web servers, databases, APIs, and any application where guaranteed delivery matters. * **UDP support** — low-latency, connectionless traffic for use cases like real-time streaming, gaming servers, DNS, VoIP, and telemetry pipelines where speed takes priority over guaranteed delivery. * **Consistent configuration experience** — TCP and UDP ports are configured the same way through the platform, keeping networking setup simple and unified regardless of the protocol your application needs. Previously, VM networking on NeevCloud was limited to TCP only. With UDP now fully supported, you can run a much broader class of workloads — including real-time data pipelines, media servers, and network tools — without workarounds or external infrastructure. For a full guide on configuring ports and networking for your VMs, see the [VM Service documentation](https://docs.ai.neevcloud.com/vm-service/). ### GPU Instances — Reserved Pricing A new **1-Month Reserved Plan** is now available for GPU instances, offering a **5% discount** compared to on-demand pricing. Designed for teams with consistent GPU utilisation, this gives you cost predictability without a long-term commitment. *** ## Known Issues ### Agentic Studio — Sandboxes * **Persistent process execution across turns:** Sandboxes currently run commands in a stateless per-request model. The ability to keep a long-running process alive inside a sandbox across multiple agent turns — such as a running server or background worker — is not yet available. This capability is on the near-term roadmap and will ship in an upcoming release. * **Outbound network access from sandboxes:** All outbound network traffic from sandboxes is currently unrestricted. A configurable egress policy, allowing you to control exactly which external endpoints your sandboxes can reach, is coming in an upcoming release. * **Billing line items:** Sandbox usage billing is active and accurate, but the format of how sandbox usage appears in your billing history may be refined in an upcoming release. You will not be charged unexpectedly. ### Agentic Studio — Agents * **Agent console:** The interactive agent console and UI are not yet available in this release. Agent management is currently API-only. The console will ship in an upcoming release. ### AI Inference — Dedicated Model Deployment * **Playground not available:** Dedicated models, including BYO models, cannot currently be tested in the Playground. This is planned for an upcoming release. * **Model pull times:** Large models may take longer than expected to pull on first deployment. We are actively working on improvements to reduce this. ### VM Service * **GPU VMs not yet available:** GPU Virtual Machines are not yet available. CPU VMs remain fully supported. GPU VM availability will be announced in a future release. ### Billing * **Usage line timing:** Usage lines for sandboxes, agents, and VM deployments may appear with a short delay after a usage period ends. This does not affect billing accuracy. * **Currency display:** Billing is processed in USD. INR and other local currency amounts shown in the platform are approximate conversions. Full local currency billing is planned for a future release. *** ## Notes June 2026 marks the beginning of NeevCloud's Agentic AI era. Sandboxes, Agents, BYO Models, a unified developer access suite, and full TCP and UDP VM networking — all shipping together — give teams everything they need to build, test, and run agentic applications on a single platform. Both Agentic Studio features — Sandboxes and Agents — are in Beta. They are fully available to all users and ready for development and early production use, with continued improvements shipping in every release. GPU VM support is on its way. We will announce availability as soon as it is ready. As always, your feedback shapes what we build next. If you run into issues or have suggestions, reach out via a support ticket or contact your NeevCloud account team. *** ## Understanding Beta Features marked as *(Beta)* are fully available to all users and ready to use, but are still being actively improved. Here is what to expect: * **Core functionality works.** Beta features are not experimental — they are production-capable, but have not yet met all the stability and completeness thresholds required for General Availability (GA). * **Frequent improvements.** You can expect updates to performance, reliability, and functionality in every release cycle. * **Some gaps remain.** Certain views or flows may not be fully complete yet. Known gaps are listed in the Known Issues section above. * **APIs may evolve.** Core endpoints are stable, but some parameters or response formats may be refined before GA. * **No full SLA coverage.** Beta features are not covered by the same uptime commitments as GA features. * **Your feedback matters.** Beta is a collaborative phase — the issues you raise and the workflows you share directly shape what we prioritise next. Please reach out. Beta features graduate to GA when they meet our internal quality, performance, and completeness thresholds. GA graduation is always announced in the release notes.