Devin Dev Stack
BeginnerDevin Desktop and Devin Cloud in one workflow, with Claude by default and other models a click away.
Published 30 September 2026
About Devin Dev Stack
Devin is Cognition's coding product, and since Windsurf became Devin Desktop it's one brand across an IDE, a cloud agent, and a CLI. Devin Desktop is a full editor built on VS Code's open-source core, but it opens on the Agent Command Center, a board where local agents running on your machine and cloud agents running on Cognition's machines sit side by side. Devin Local, the agent inside the editor, plans multi-file changes, runs terminal commands, and can split work across subagents.
Models come from Devin's own picker. Claude is the default here, with OpenAI's GPT models, Gemini, and Grok in the same picker, and several can be in use at once across different sessions. Cognition's own SWE-2 model sits alongside them, included on paid plans; it's built for Devin and isn't available anywhere else.
Work keeps going away from the desk through Devin Cloud, included from the Pro plan: a cloud session runs on Cognition's infrastructure with your machine off and hands back a pull request. Sessions start from the Devin web app, the Agent Command Center, a mention in Slack, Microsoft Teams, Linear, or Jira, or a comment on a GitHub pull request, which makes it easy to hand over a ticket without opening the editor. There's no released mobile app yet: an iOS app is in a waitlist beta, and the web app works in a phone browser.
A separate coding agent is an optional extra: on a paid plan, the Agent Client Protocol (still in preview) lets Claude Code, Codex, or OpenCode run in the same interface as Devin's own agent, billed by their own providers. Devin also has its own pull-request reviewer, Devin Review, so a third-party AI reviewer is optional too. Repositories live on GitHub by default or on GitLab for self-hosted source control, and Docker or a CI/CD pipeline are common additions; neither is needed to start.
Key Features
- ✓Devin Desktop editor on VS Code's open-source core, with extensions and keybindings carried over
- ✓Agent Command Center managing local and cloud agents on one board
- ✓Devin Cloud sessions that keep working with your machine off and return pull requests
- ✓Cloud sessions started from Slack, Teams, Linear, Jira, GitHub, or the web app
- ✓Claude by default, with GPT, Gemini, Grok, and Cognition's SWE-2 in the same picker
- ✓Claude Code, Codex, or OpenCode in the same interface through the Agent Client Protocol on paid plans
When to Use Devin Dev Stack
- →Handing a Linear or Jira ticket to a cloud session and reviewing the pull request later
- →Running a local agent on one task while cloud agents work through others
- →Switching between Claude and a faster model per session inside one editor
- →Moving from Windsurf or VS Code without losing extensions or keybindings
- →Trying Devin's local agent on the free tier before adding cloud sessions
Pros
- Local and cloud agents managed from one board instead of two separate products
- Cloud sessions start from the tools a team already uses, like Slack, Linear, and Jira
- Model choice across Claude, GPT, Gemini, Grok, and SWE-2 from one subscription
- Built on VS Code's open-source core, so existing extensions and keybindings carry over
Cons
- Devin Cloud needs the $20 Pro plan; the free tier is Desktop-only with limited models
- No released mobile app for following cloud sessions yet, though the web app works in a phone browser and an iOS app is in beta
- Heavy cloud use runs into plan quotas, with on-demand credits beyond them
- The Windsurf-to-Devin rebrand means older guides and tutorials use outdated names
LLM Options for Devin Dev Stack
Devin Dev Stack powered by Claude
Claude is the default model here: a common pick for large multi-file changes and careful reasoning about unfamiliar code, available in Devin's picker on paid plans without a separate Anthropic subscription. Cognition's own SWE-2 sits beside it as the model built specifically for Devin.
Devin Dev Stack powered by OpenAI
OpenAI's GPT models are in the same picker, useful for a second opinion when one model gets stuck, or for teams already standardized on OpenAI. They're covered by Devin's plan quotas, with on-demand credits beyond them, so there's no separate OpenAI account to manage.
Devin Dev Stack powered by Google Gemini
Gemini's Flash models are a fast, lower-cost pick for everyday edits and quick agent loops, stretching Devin's plan quotas further than the premium models, and their long context window suits repository-wide questions in Devin Desktop or a cloud session.
Devin Dev Stack powered by Grok
Grok models, from SpaceXAI (formerly xAI), round out the picker, another fast option for routine work when you want to compare results across providers. Like the other frontier models, they're available on Devin's paid plans and draw on the same quota.
These are highlighted picks. To see all the tools, check the LLM category.
Version Control Options for Devin Dev Stack
Devin Dev Stack with GitHub
GitHub is the default here: the largest pull-request ecosystem, and Devin can open pull requests, respond to review comments on them, and start a cloud session from a comment on an open pull request.
Devin Dev Stack with GitLab
GitLab works with Devin Cloud too, and bundles source control, merge requests, and CI/CD in one self-hostable platform for teams whose code has to stay on their own infrastructure.
Devin Dev Stack Add-ons
Each addition below extends this stack with a capability the base stack works fine without. None are required: include the ones your product actually needs when building this stack, and skip the rest.
Coding Agent Add-ons
Add a separate coding agent when you want another vendor's agent working alongside the one built into the editor, either through its own extension or in the integrated terminal.
Devin Dev Stack with Claude Code
Add Claude Code when you want Anthropic's agent working alongside Devin's own. On a paid Devin plan it runs in Devin Desktop's interface through the Agent Client Protocol, uses Claude models only, and bills through its own Claude plan, from Pro at $20 a month.
Devin Dev Stack with Codex
Add Codex when you want OpenAI's agent on a task inside Devin Desktop. On a paid Devin plan it connects through the same Agent Client Protocol, uses OpenAI's models by default, and bills through a ChatGPT plan you may already have, with a free tier and Plus at $20 a month.
Devin Dev Stack with OpenCode
Add OpenCode when you want an open-source agent in Devin Desktop that takes almost any provider's model. On a paid Devin plan it connects through the Agent Client Protocol, usually with your own API key and per-token billing from that provider, which can be cheaper with open-weight models.
These are highlighted picks. To see all the tools, check the AI Coding Agents category.
Code Review Add-ons
Add code review when you want an AI reading every pull request before it merges: it comments on the diff so bugs, security issues, and inconsistencies surface before a human has to catch them.
Devin Dev Stack with CodeRabbit
Add CodeRabbit when you want an AI reviewer on every pull request: it installs as a GitHub or GitLab app, summarizes the diff, and leaves line-level comments before anyone merges, with a free tier for open-source repos.
Devin Dev Stack with Greptile
Add Greptile when you want review that reads the whole repository rather than just the diff: it indexes the codebase so its GitHub and GitLab pull request comments carry wider context, and self-hosting is available on its Enterprise plan.
Devin Dev Stack with Cursor Bugbot
Add Cursor Bugbot when a bug-focused reviewer is enough: it reviews GitHub or GitLab pull requests whatever wrote the code, though it bills through a Cursor account and plan rather than its own subscription.
CI/CD Add-ons
Add CI/CD when you want a dedicated pipeline for running tests, linting, or multi-stage builds before a deploy goes out. Many hosting platforms already redeploy automatically on every push on their own — a CI/CD tool adds the most value on top of that by gating the deploy on a passing test suite, and matters even more when the hosting choice does not auto-deploy at all, such as a self-hosted server.
Devin Dev Stack with GitHub Actions
Add GitHub Actions when Devin's cloud sessions open pull requests on their own and every one should be tested automatically before you review it.
Devin Dev Stack with GitLab CI/CD
Add GitLab CI/CD for the same automatic checks when the repository lives on GitLab, with pipelines that can run on your own runners.
These are highlighted picks. To see all the tools, check the CI/CD Pipelines category.
Containerization Add-ons
Add containerization when you want the app packaged the same way across local development, staging, and production, or need to deploy somewhere that isn't a managed serverless platform.
Devin Dev Stack with Docker
Add Docker when the project needs the same environment on a laptop, in Devin's cloud sessions, and in production, so an agent's passing tests mean the same thing everywhere.
These are highlighted picks. To see all the tools, check the Containerization category.
Frequently Asked Questions about Devin Dev Stack
Can Devin keep working when my computer is off?
Yes, with Devin Cloud, included from the $20 Pro plan. A cloud session runs on Cognition's own infrastructure with your machine off and hands back a pull request. You can start one from the Devin web app, the Agent Command Center in Devin Desktop, by mentioning Devin in Slack, Microsoft Teams, Linear, or Jira, or by commenting on a GitHub pull request. Local sessions in Devin Desktop stop when your computer sleeps. There's no released mobile app yet (an iOS app is in a waitlist beta), so checking in from a phone means the web app in a browser or a Slack thread.
Which models can Devin use, and what is SWE-2?
Devin's picker covers Claude, OpenAI's GPT models, Gemini, Grok, and a few open-weight models, with full availability on paid plans; the free tier offers a limited set. SWE-2 is Cognition's own coding model, built for Devin and only available inside it. Claude is this stack's default, but models are chosen per session, so mixing them is normal.
How is this different from using Cursor?
Both are VS Code-based editors with their own agent, a multi-provider model picker, cloud agents, and Tab completions. Devin opens on the Agent Command Center, a board that lists local and cloud sessions together, and its cloud sessions start from Jira and Microsoft Teams as well as Slack, Linear, and GitHub. Cursor has a native iOS app for following its cloud agents, while Devin's own iOS app is still in a waitlist beta. Pick Devin to hand tickets from a tracker to cloud sessions and watch all your agents on one board; pick Cursor if a native phone app for cloud agents matters.
What does heavy use cost?
Everything Devin does draws on one quota: Devin Desktop, the CLI, and cloud sessions share it. Pro has a daily and a weekly allowance; Max has a larger weekly one with no daily cap. Each model uses the quota at its own token price, so a lighter model such as Gemini Flash goes further than a frontier one on routine edits. Past the allowance, on-demand credits bill at API rates, and purchased credits don't expire. On Pro, Max, or Teams you can link an eligible ChatGPT plan so GPT usage bills there instead of Devin's quota. An agent added through the Agent Client Protocol bills separately with its own provider. The Pricing section has the plan figures.
Can I use Claude Code or Codex inside Devin?
Yes, on a paid plan. Devin Desktop runs third-party agents through the Agent Client Protocol, which is still in preview, so Claude Code, Codex, and OpenCode can sit in the same interface as Devin's own agent. Each one uses its own provider's models rather than Devin's picker: Claude Code runs Claude, Codex defaults to OpenAI's models, and OpenCode takes almost any provider with your own key. Billing and terms are directly with that agent's provider, and Devin's subscription only covers its own agents and models.
Stacks Related to Devin Dev Stack
Paseo Dev Stack
Developer
Run Claude Code, Codex, and other coding agents on your own machine with Paseo and steer them from any device.
Claude Code Dev Stack
DeveloperClaude Code in the terminal, desktop app, or web, paired with Claude and the add-ons to run it anywhere.
T3 Code Dev Stack
Developer
One open-source app for Claude Code, Codex, OpenCode, and other coding agents, on desktop and mobile.
Conductor Dev Stack
Developer
Run Claude Code, Codex, and Cursor's agent in parallel on your Mac with Conductor, each in its own workspace.
Scores
Popularity4/5
Windsurf and Devin were each widely known, and the combined Devin brand still trails Cursor and VS Code in daily use.
Learning Curve2/5
A VS Code-based editor feels familiar at once, but getting the most out of it means learning to delegate through the Agent Command Center and cloud sessions.
Flexibility4/5
Several model providers, local and cloud agents, ACP for third-party agents, and integrations from Slack to Jira, within Cognition's closed product.
Performance4/5
Devin Local is quick for in-editor work and cloud sessions move long tasks off the machine, with plan quotas as the practical ceiling.
Portability4/5
Projects stay plain files and git history on a VS Code base, so moving to another editor costs little beyond Devin-specific workflows.
Tools in the Devin Dev Stack Stack
Development Tools
Add-ons (optional — add any, or none)
Coding Agent
Code Review
CI/CD
Containerization
Devin Dev Stack Pricing
Devin's subscription is the main cost: the free tier covers Devin Desktop with light quotas and limited models, Pro at $20/month adds Devin Cloud and the full model picker, and Max at $200/month raises quotas for heavy use. Teams has an $80/month minimum, with full seats at $40 each. A separate agent like Claude Code or Codex needs its own plan. GitHub and GitLab are free for individual developers, and Docker and CI/CD have free tiers.
Free covers Desktop only; Pro $20/mo adds Devin Cloud and all models; Max $200/mo; Teams $80/mo minimum with $40 per full seat.
Free tier covers individual devs; GitHub Team is $4 per user per month, GitLab Premium is $29 per user per month.
Claude Code starts at $20/mo (no free tier); Codex has a free tier, Plus $20/mo; OpenCode is free and bills per token.
Docker Personal is free; Docker Pro is $11/mo for individual professionals.
GitHub Actions and GitLab CI/CD both include free monthly minutes before paid tiers kick in.