AI Agent Index
ByHeather MacAvelia·Independently reviewed·Published Mar 18, 2026·Updated Sep 20, 2026
Independently verified against live vendor data on Sep 7, 2026.

AI-powered IDE from Cognition, formerly Windsurf, pairing a local editor with asynchronous cloud agents and MCP in both directions. Free tier, Pro $20 per month, Max $200 (USD).

How we scored it

Autonomy

4/5

Integrations

5/5

Pricing clarity

4/5

Evidence

4/5

Setup

4/5

The facts

From

$20

freemium ↗

GitHub

—

Stars

G2

4.3 / 5

35 reviews ↗

MCP

⚡ Server + client

Server + client

Reviews

Be the first →

Devin Desktop, formerly Windsurf, is the AI-powered IDE from Cognition. The rename landed on June 2, 2026 as a standard over-the-air update. The Agent Command Center is now the primary view: Spaces share context and Git worktrees across agents, a Kanban board tracks running sessions, and local and cloud agents are managed from one surface. The full Windsurf editor sits underneath with its extensions, keybindings and workflows intact, and a separate Windsurf for JetBrains plugin remains available for download. Agents from other vendors run inside the editor through the Agent Client Protocol (ACP). Cognition describes engineers reviewing every agent diff before pushing and, on large migration work, approving the agent's changes, so this is autonomous execution under review. The plan comparison names Slack, Microsoft Teams, Linear, Jira, GitHub, GitLab and Bitbucket, with custom git provider support on higher tiers. Devin Desktop is an MCP client, and its product page lists nine MCP servers a team can connect: Slack, Linear, Notion, Figma, Sentry, Stripe, Vercel, Datadog and Atlassian for Jira and Confluence. Cognition also publishes its own MCP server at mcp.devin.ai, documented on docs.devin.ai, which gives any MCP client access to session management, playbooks, knowledge and scheduling on the Devin platform. A public Devin API is included from the Pro tier. Pricing is public, in USD, and shared with the Devin platform rather than sold separately. Free at $0 includes a light agent quota, limited model availability, unlimited inline edits and unlimited Tab completions. Pro at $20 per month adds increased quotas with OpenAI, Claude and Gemini frontier models, full model availability, free use of SWE 1.7 and access to Devin Cloud. Max at $200 per month raises quotas substantially. Teams is $80 per month for the plan plus $40 per month per full developer seat. A full seat is what unlocks Devin Desktop on Teams, along with $40 of monthly usage that works across Devin Cloud, Devin Desktop, Devin CLI and Devin Review. Enterprise is quote-only. Usage allowances refresh daily and weekly, and a paid plan that exceeds its allowance can buy extra usage at API pricing. No annual rate is published on any surface and the pricing page carries no billing toggle. What the entry price does not buy is specific. On the Free plan, premium models, DeepWiki, the Devin API, Ask Devin and the advanced capabilities row are all switched off, and SWE 1.7 runs at a slower rate. SAML and OIDC single sign-on, centralized enterprise admin controls, teamspace isolation and virtual private cloud deployment are Enterprise-tier items, and concurrent sessions are capped at ten below the Teams tier. Under the Cognition Platform Terms, customer data may be used for model training by default, the contractual opt-out is available only on paid tiers, and on Teams only an administrator can exercise it. Customers on a legacy Windsurf Master Services Agreement are governed differently: that contract states customer data will not be used to train any model without written consent. Cognition's Trust Center runs two scopes. The Cognition scope shows CCPA, SOC 2 Type 2 and ISO/IEC 27001:2022. The Cognition Public Sector scope states that local agents, meaning Devin Desktop, are FedRAMP High authorized today and meet DoD IL4, IL5 and IL6 and ITAR, with FIPS 140-2 and 140-3, NIST 800-53 Rev. 5 and 800-171 Rev. 3 and a CMMC Level 1 self-assessment covering those surfaces, delivered through Palantir FedStart. The FedRAMP Marketplace lists that boundary under Palantir Federal Cloud Service Supporting Services as FedRAMP Certified, Class D (High). The Cognition Platform cloud agents are listed there as Agency Authorization In Process and are slated for FedRAMP High in the second half of 2026. The underlying reports on both scopes sit behind an access request and an NDA. Developers who want a lighter interactive assistant without the autonomous cloud layer should evaluate Cursor at $16/mo billed annually or GitHub Copilot at $10/mo for broad editor support. Claude Code at $17/mo billed annually offers terminal-native agentic coding with no dedicated IDE. Current state Q3 2026: Devin Desktop is the shipping name and windsurf.com now resolves to the Devin site, with legacy Windsurf plans still managing settings there. The G2 profile remains under the Windsurf name and the Exafunction entity. Gartner named Cognition (Windsurf) a Leader in the 2025 Magic Quadrant for AI Code Assistants in September 2025. Ramp, Harvey and NVIDIA are named as teams building with Devin Desktop, with NVIDIA in the multi-agent research preview. Models available inside the editor include OpenAI, Claude and Gemini frontier models, Cognition's SWE 1.7, and open source models such as Kimi K2.5.

Pricing

freemium · $20/mo

View pricing ↗

Segment

b2b

Setup

easy

Verified

Sep 7, 2026

Transparency

Mostly Public

Contract

Month-to-month

Data training

Opt-out

Human in loop

Optional

Capabilities

ideagentic-codingmulti-file-editingautonomousterminal-agentcode-generation

Pros & Limitations

Editorial assessment

Pros

  • ✓Bundles interactive IDE coding with asynchronous cloud agents under one subscription, so a developer can keep editing locally while delegating longer tasks to Devin Cloud sessions and reviewing the results in the same surface.
  • ✓MCP works in both directions, which is rare in this category. Devin Desktop connects out to external MCP servers such as the Slack MCP Server, and Cognition publishes its own MCP server with documented setup instructions for streamable HTTP clients.
  • ✓Pricing is public and self-serve up to the Teams tier, with a permanently free plan and a documented usage-quota model, so a team can evaluate and adopt the product without talking to sales.
  • ✓Devin Desktop is FedRAMP High authorized and meets DoD IL4, IL5 and IL6 through Palantir FedStart, with ITAR and FIPS 140-2 and 140-3 coverage on the same scope.

Limitations

  • ⚠SAML and OIDC single sign-on, virtual private cloud deployment, teamspace isolation and centralized admin controls are all Enterprise-tier items, so the $20 Pro plan carries none of them and concurrent sessions are capped at ten.
  • ⚠Customer data may be used for model training by default under Cognition's platform terms, and the contractual opt-out is restricted to paid tiers, so free-plan users have no opt-out at all. On the Teams tier only an administrator can exercise it.
  • ⚠Cognition publishes customer outcome figures against anonymized categories such as Financial Institution or Technology Company rather than named accounts, and publishes no autonomous task-completion rate on any surface. FedRAMP High and the DoD impact levels cover Devin Desktop today, while the Devin Cloud agents are slated for the second half of 2026.

Technical Details

Deployment
desktopclicloudide
Model architectureMulti-model. SWE 1.7 proprietary plus OpenAI, Anthropic and Google frontier models and open-source models.
Autonomous rateCognition documents autonomous multi-step execution by local and cloud agents, with engineers reviewing agent diffs before pushing and, on large migration projects, approving the agent's changes. The customers page publishes outcome multiples such as a 70% reduction in .NET migration time attributed to anonymized customer categories rather than named accounts, and no autonomous task-completion rate is published on any Cognition surface as of Q3 2026.
Integrations
GitHubGitLabBitbucketSlackMicrosoft TeamsLinearJiraVS CodeJetBrainsNotionFigmaSentryStripeVercelDatadogConfluence
Security
SOC 2 Type IIISO 27001CCPAFedRAMP

Similar agents

Rating

4.3/ 5
★★★★★

Editorial score

How we score this →

Recognition

MCP Server VerifiedListed 2026
Embed these on your site →

Industries

Public SectorLegalFinance

Built for

DevToolsStartupsSaaSB2BEnterprise

Leave a review

Never displayed publicly.

Agent Stacks

See workflow stacks that feature Windsurf.

Compare

Related Content

Is this your tool?

Claim this listing free to get your embeddable badges, a review collection link, and the Claimed badge on your listing.

Claim listing →

Free · Every Two Weeks

AI Agent Price & Rating Tracker

Price changes, new agent launches, acquisitions, and rating updates across the AI agent market. Verified against live vendor data, not vendor marketing.

No spam. Unsubscribe anytime. We never share your email.