Gitpod is now Ona
The first fact a 2026 buyer must understand is that Gitpod is now Ona. The company announced the rebrand on September 2, 2025 and repositioned itself as mission control for software engineering agents across the development lifecycle. Ona still provides sandboxed development environments and a full VS Code experience in the browser, but the center of gravity has moved from typing code in a cloud IDE to delegating, supervising, and reviewing work performed by agents. A review that treats the old Gitpod product and today's Ona platform as unchanged would mislead readers about both the workflow and the commercial decision.
This was not merely a new logo attached to Gitpod Classic. Before the company rename, Gitpod had already introduced a replatformed architecture, first called Flex and then simply Gitpod. The official migration note distinguished app.gitpod.io as the new architecture and gitpod.io as Gitpod Classic, with Classic scheduled to sunset in September 2025. Older claims about 50 free workspace hours, the Classic dashboard, or the previous SaaS behavior should therefore be treated as historical. Current evaluation belongs on Ona's agent sessions, environments, organization controls, and OCU-based usage.
What the product does now
Ona combines AI agent conversations with disposable or persistent development environments that contain the repository, tools, terminals, and preview services needed to complete software tasks. The company describes the product as a personal team of autonomous software engineers, while preserving full VS Code in the browser and access from a phone for supervision. This makes the platform broader than an autocomplete assistant: the agent works inside an isolated environment, can plan and modify code, and can return pull requests, preview links, or other artifacts for human review.
The environment layer remains strategically important because agents need the same runtimes, dependencies, secrets, and network access as human developers. Ona supports project configuration, prebuilds, environment lifecycle management, GPU-capable classes, and parallel environments. The buyer value is highest when background work can be separated into multiple reproducible workspaces without overloading a laptop. A team that only wants a remote editor can still use the environment, but it would be paying for and operating inside a product whose roadmap and interface are explicitly organized around agentic engineering.
Pricing and OCU economics
Ona's current Core plan starts at $20 per month and uses Ona Compute Units, or OCUs, for both environment runtime and agent conversations. The pricing page lists Core packages with 80 to 2,200 recurring monthly OCUs and add-ons starting at $10 for 40 OCUs. A standard environment with 4 vCPUs and 16 GB RAM consumes 1 OCU per hour, while a GPU environment with 16 vCPUs and 64 GB RAM consumes 7 OCUs per hour. Ona's examples estimate roughly 1 OCU to explain a small codebase, 4 to create a new web app, and 8 to add a feature to a medium codebase, while warning that real agent usage varies.
That blended meter is convenient but requires a different budget model from old cloud-IDE hours. Subscription OCUs renew monthly and do not roll over; top-up credits can remain valid for up to one year while a subscription is active. Core organizations can enable automatic top-ups, so finance owners should decide whether uninterrupted agent work or a hard spending boundary matters more. The strongest procurement test is a representative sprint: estimate standard-environment hours, GPU hours, and the mix of agent tasks, then compare that OCU demand with the package rather than assuming the $20 entry price covers sustained team use.
Governance, privacy, and lifecycle
Core supports up to 100 team members without per-seat pricing, unlimited parallel environments, role-based access management, project secrets, custom organization commands, and deny lists. It can also connect an existing Codex subscription so model usage is covered by that subscription while Ona bills for environments and compute. Ona states that customer code and data are not used to train models. On Core, environments run in Ona's multi-tenant AWS deployment and inactive environments are automatically deleted after seven days, with reminders intended to prompt users to commit remaining work.
Enterprise changes the control plane materially: Ona lists customer-VPC deployment, network control, SSO and OIDC, organization-wide secrets, detailed audit trails, MCP controls, SDK and API access, and custom lifecycle policy. Those features make Enterprise the relevant comparison for organizations that previously valued self-hosted Gitpod. Core buyers should treat the seven-day deletion rule as an operational requirement, not a footnote, and make commit or artifact persistence part of every agent workflow. Neither agent autonomy nor an isolated VM substitutes for branch protection, review requirements, scoped credentials, or audit retention.
Verdict and migration choices
Ona is a strong candidate for teams that want agents and development environments purchased as one operational platform. It is most persuasive when work can be divided into parallel, reviewable tasks and when developers already accept remote environments as the execution layer. The ability to connect Codex, run full VS Code, choose large or GPU-backed machines, and pool OCUs across an organization gives Core unusual flexibility for small teams. The tradeoff is that usage forecasting now spans model-driven tasks and infrastructure, and the product is intentionally moving beyond the conventional cloud-IDE category.
Teams migrating from Gitpod Classic should not renew an old mental model. They should inventory Classic workspaces, confirm that repositories and uncommitted changes are preserved, rebuild environment configuration on Ona, and validate secret, network, and compliance requirements against the current tier. GitHub Codespaces is the cleaner choice for a GitHub-only remote workstation; Coder is stronger when self-hosted neutrality is essential; local devcontainers remain simpler for teams that do not need background agents. Ona wins when agent orchestration plus managed environments is the requirement, not when the goal is to recreate Gitpod Classic exactly.
