Quick Verdict: Gitpod vs Codespaces
Bottom Line
These tools no longer compete directly. Gitpod rebranded to Ona in September 2025 and pivoted to AI software engineering agents. GitHub Codespaces is now the dominant cloud dev environment for GitHub-hosted repos. If you need a CDE, Codespaces wins by default. If you need AI agent infrastructure, Ona is worth evaluating alongside Morph and Devin.
The Ona Rebrand: Why Gitpod Left CDEs Behind
Gitpod built its reputation on ephemeral, prebuilt cloud workspaces. You opened a GitHub repo link prefixed with gitpod.io/# and got a running dev environment in seconds. It supported GitHub, GitLab, and Bitbucket, used gitpod.yml for configuration, and ran containers on Kubernetes.
By mid-2025, the company concluded that CDEs were a shrinking market. AI coding agents don't need browser-based IDEs. They need sandboxed execution environments with API access, security guardrails, and orchestration. Gitpod deprecated its Classic SaaS product, shipped Gitpod Flex as a self-hosted CDE bridge, and relaunched as Ona in September 2025.
The bet: the infrastructure that ran developer workspaces is the same infrastructure AI agents need. Isolated containers, fast provisioning, secure network boundaries. Ona repurposed Gitpod's container orchestration layer for agent sandboxing rather than human IDE sessions.
What Happened to Gitpod Classic?
Gitpod Classic pay-as-you-go was sunset on October 15, 2025. Existing workspace data is no longer accessible through the old SaaS. Teams still wanting Gitpod-style CDEs can use Gitpod Flex (self-hosted) or migrate to Codespaces, Coder, or DevPod.
Feature Comparison: Gitpod Classic vs Codespaces
For historical context and teams still evaluating CDE options based on older comparisons, here is how the two products stacked up before Gitpod's pivot.
| Feature | Gitpod Classic | GitHub Codespaces |
|---|---|---|
| Git host support | GitHub, GitLab, Bitbucket | GitHub only |
| Editor | VS Code (browser), JetBrains | VS Code (browser + desktop) |
| Config format | gitpod.yml | devcontainer.json |
| Prebuilds | Yes (automatic) | Yes (manual or GitHub Actions) |
| Workspace lifecycle | Ephemeral by default | Persistent, suspendable |
| Self-hosted option | Yes (Gitpod Flex) | No |
| GPU support | No | Limited (select machine types) |
| Regions | EU, US (GCP-based) | 4 regions (US West, US East, EU West, SE Asia) |
| Container runtime | Kubernetes pods | Azure VMs |
| Free tier | 50 hours/month (was) | Included with GitHub account |
| Status (2026) | Deprecated (SaaS sunset) | Active, expanding |
GitHub Codespaces in 2026
Codespaces has strengthened its position as the default CDE through Microsoft's vertical integration. VS Code, GitHub, Copilot, GitHub Actions, and Codespaces form a single stack where each layer feeds the next.
30-Second Spin-Up
A fully configured dev environment, including Docker-in-Docker setups, launches in about 30 seconds. Prebuilds via GitHub Actions eliminate cold starts for frequently used repos.
Native Copilot Integration
Copilot runs as a first-party extension inside Codespaces. Code completion, chat, and agent mode work without additional API keys or configuration. The AI assistant has full access to the workspace context.
devcontainer.json Standard
Codespaces uses the open Dev Container spec, which is also supported by VS Code locally, JetBrains, and other tools. Your CDE config is portable, not locked to GitHub.
Machine Types Up to 32 Cores
Choose from 2-core to 32-core machines. Storage at $0.07/GiB/month. Suspended instances cost nothing for compute. Billing is transparent and per-minute.
The main limitation: Codespaces only works with GitHub repositories. Teams using GitLab or Bitbucket need Coder, DevPod, or a self-hosted solution.
Ona (Formerly Gitpod) in 2026
Ona is no longer a CDE in the traditional sense. It is an AI agent platform built on top of Gitpod's container infrastructure. The product has three layers: Ona Environments (sandboxed dev containers), Ona Agents (autonomous coding agents), and Ona Guardrails (RBAC, audit trails, SSO, command-level controls).
Assisted + Autonomous Modes
Assisted mode keeps the developer in control with agent-powered completions. Autonomous mode lets agents execute entire tasks end-to-end: planning, coding, testing, and creating PRs inside secure environments.
Enterprise-Grade Guardrails
VPC deployment, fine-grained RBAC, SSO/OIDC, org-wide secrets, and audit trails for every agent action. Agents run in kernel-level sandboxes with network controls.
Multi-Provider LLM Support
Ona connects to Anthropic, Google Vertex, AWS Bedrock, and other providers. Enterprises pick the LLM that meets their compliance and performance requirements.
Ona Compute Units (OCUs)
Billing is measured in OCUs, which bundle environment provisioning and agent compute. Creating a new web app from scratch: ~4 OCUs. Adding a feature to a medium project: ~8 OCUs.
Internally, Ona claims its agents co-authored 60% of merged PRs and contributed 72% of code merged to main in a representative week. These numbers come from Ona's own engineering team, so take them as a signal of how aggressively the company dogfoods rather than as a universal benchmark.
Pricing
Direct price comparison is difficult because these products now solve different problems. Codespaces charges for compute time and storage. Ona charges in OCUs that bundle environment and agent costs.
| Item | GitHub Codespaces | Ona |
|---|---|---|
| Unit of billing | Per-minute compute + storage | Ona Compute Units (OCUs) |
| Starting price | $0.18/hr (2-core) | Free tier available; Enterprise is custom |
| Storage | $0.07/GiB/month | Included in OCU |
| Free tier | Yes (included with GitHub account) | Yes (limited OCUs) |
| Enterprise pricing | Per-seat GitHub Enterprise + usage | Custom (contact sales) |
| AI costs | Copilot subscription ($10-39/mo) | Included in OCU (BYOL for enterprise) |
For pure CDE usage, Codespaces is cheaper and more predictable. A developer running a 4-core instance 8 hours/day for 20 working days pays roughly $57.60/month in compute. Ona's OCU model makes more sense when you factor in agent execution, where the value is measured in tasks completed rather than hours billed.
When Codespaces Wins
GitHub-Native Teams
If your repos, CI/CD, project management, and code review all live on GitHub, Codespaces eliminates environment setup entirely. One click from a PR to a running workspace.
Onboarding and Education
New developers clone nothing. They open a codespace and start coding. Universities and bootcamps use Codespaces to eliminate the 'works on my machine' problem for every student.
Copilot-First Workflows
Copilot's agent mode inside Codespaces has full workspace context: files, terminal, dependencies. No extra setup. For teams whose AI usage is Copilot-centric, the integration is seamless.
Predictable Billing
Per-minute compute, per-GiB storage. No opaque unit pricing. Suspended instances cost nothing. Finance teams can model CDE spend accurately from usage data.
When Ona Wins
AI Agent Infrastructure
If your goal is running autonomous coding agents, not human developers in browser IDEs, Ona's sandboxed environments and guardrails are purpose-built. Codespaces was designed for people.
Enterprise Compliance
VPC deployment, kernel-level isolation, command-level audit trails, and RBAC for agent actions. If your security team needs to approve every tool an AI agent can call, Ona has the controls.
Multi-Git-Host Teams
Ona's agent platform works across GitHub, GitLab, and Bitbucket. Codespaces is GitHub-only. For organizations with repos spread across multiple hosts, Ona provides unified agent coverage.
Agent-First Development
Ona's autonomous mode handles planning through deployment. If you want agents writing 50%+ of your code in sandboxed, auditable environments, that is the product's core promise.
AI Agents Need Sandboxes
Ona's pivot reflects a real trend. AI coding agents that execute code, run tests, and modify filesystems need isolated environments. Giving an autonomous agent access to your local machine or a shared server is a security risk. The agent needs its own sandbox with controlled network access, resource limits, and audit logging.
This is why CDE infrastructure maps well to agent infrastructure. Both need fast container provisioning, workspace isolation, and programmatic lifecycle management. The difference is the user: a human with a keyboard vs. an LLM with tool-use capabilities.
Codespaces solves this partially. You can run a Copilot agent inside a codespace. But the environment is optimized for human interaction (VS Code UI, port forwarding, terminal access) rather than agent throughput (API-driven, headless, parallelized).
Morph for AI Agent Sandboxes
Morph provides fast, API-driven execution environments for AI coding agents. If you're building or deploying agents that need sandboxed code execution, file editing, and test running, try Morph for agent-optimized infrastructure with sub-second provisioning and full API control.
Frequently Asked Questions
Is Gitpod still available as a cloud dev environment in 2026?
No. Gitpod Classic SaaS was deprecated in 2025. The company rebranded to Ona and pivoted to AI software engineering agents. Gitpod Flex exists as a self-hosted CDE option, but the company's development focus is on agent infrastructure. Teams needing a hosted CDE should evaluate GitHub Codespaces, Coder, or DevPod.
What is Ona (formerly Gitpod)?
Ona is the rebranded Gitpod, launched September 2025. It runs autonomous AI coding agents in sandboxed cloud environments with enterprise guardrails (VPC, RBAC, audit trails). It offers assisted mode where developers stay in control and autonomous mode where agents handle tasks end-to-end. The platform uses Ona Compute Units for billing.
How much does GitHub Codespaces cost?
Compute starts at $0.18/hour for 2-core machines, scaling to $2.88/hour for 32 cores. Storage costs $0.07/GiB/month. All GitHub accounts include a free monthly quota. Suspended instances are not billed for compute. A developer on a 4-core machine for a full working month pays roughly $58.
Should I switch from Gitpod to GitHub Codespaces?
If you need a cloud development environment for human developers, yes. Codespaces is the strongest option for GitHub-hosted repos with 30-second spin-up, native Copilot integration, and the open devcontainer.json standard. For repos on GitLab or Bitbucket, look at Coder or DevPod.
Can I use Gitpod/Ona with GitLab or Bitbucket?
Legacy Gitpod supported all three Git hosts. The Ona agent platform still integrates with multiple Git providers for its AI workflows. GitHub Codespaces only works with GitHub repositories. Multi-host support was always Gitpod's advantage and remains one for Ona in the agent context.
Related Comparisons
AI Agent Sandboxes, Built for Speed
Morph provides fast, API-driven execution environments for AI coding agents. Sandboxed code execution, sub-second provisioning, and full programmatic control. Try it free.