agentboards.org
Compare/Google Antigravity vs Windsurf (Devin Desktop)

Google AntigravityvsWindsurf (Devin Desktop)

Generated from the two spec rows. Green marks the better value where a spec has a clear direction. Everything else is just different.

Google Antigravity
Google · AI-native IDE
#68MCP
Panel
6.0
4 spec wins
Reliability
5.3
Usefulness
6.7
Cost
6.2
Longevity
5.8

“Google's own IDE ships Claude and gpt-oss next to Gemini, the most honest model review Google has ever published.”

Windsurf (Devin Desktop)
Cognition · AI-native IDE
#146MCP
Panel
5.2
0 spec wins
Reliability
5.7
Usefulness
5.8
Cost
4.8
Longevity
4.3

“Bought in July 2025, renamed in June 2026; the editor is stable, the name is not.”

Spec by spec

SpecGoogle AntigravityWindsurf (Devin Desktop)
Architecture
CategoryAI-native IDEAI-native IDE
Runslocallocal, cloud
Platformsmacos, linux, windowsmacos, linux, windows
Context windownot documentednot documented
Protocols
MCP clientYesYes
MCP serverNoNo
Capabilities
Runs terminal commandsYesYes
Multi-file editsYesYes
Git operationsYesYes
Browser controlYesA first-party browser subagent drives Chrome from a separate profile with allowlist/denylist controls (https://antigravity.google/docs/ide/browser/); IDE-side only, not in the agy CLI. YesPreviews open in a built-in browser pane next to the agent, and selected elements plus captured console errors are sent back to it as context; Cascade itself only adds web search.
Sandboxed executionYesOS-level isolation rather than containers: "no virtual machines or Docker images to manage" - Linux uses kernel namespaces to isolate the filesystem, hide host processes and cut networking; macOS uses sandbox-exec with Seatbelt (SBPL) profiles. YesThe Devin Local agent, the shared Devin CLI harness inside Devin Desktop, supports OS-level sandboxing with filesystem isolation and network filtering (docs.devin.ai/desktop/devin-local).
Multi-agent orchestrationYesYes
Headless / CI modeYesNoDevin Desktop is an IDE with no documented non-interactive or CI run mode; scripted runs belong to the separate Devin CLI and cloud API.
Models
BackboneGemini 3.1 Pro, Gemini 3.8 Flash, Claude Sonnet 4.6, Claude Opus 4.6, gpt-oss-120bSWE-1.7, Claude, GPT, Gemini, Grok, DeepSeek, Kimi, GLM
Bring your own modelYesSetting `modelProvider: "gemini"` in ~/.gemini/antigravity-cli/settings.json plus a GEMINI_API_KEY runs the CLI on your own key and billing instead of a signed-in Google account, and a model selector picks between Gemini, Claude and gpt-oss models in the IDE. NoModels are Cognition-hosted only — no custom API key, gateway, Bedrock, Vertex or Azure deployment option is documented for Devin Desktop.
Local modelsYesGOOGLE_GEMINI_BASE_URL sends model requests to any Gemini-compatible endpoint, so a self-hosted shim works as a mechanism; no Ollama or LM Studio integration is documented. No
Cost
Pricing modelmixedmixed
Starts at$20/mo$20/mo
Free tierYesYes
Bring your own keyYesNo
Openness
Open sourceNoNo
Licenseproprietaryproprietary
GitHub stars2,446n/a

Which one would each critic pick

CriticGoogle AntigravityWindsurf (Devin Desktop)Pick
El Juez——not enough reviews
El Amigo7.06.3Google Antigravity — Pick Antigravity if you want several agents running at once with a browser checking their work; pick Cursor if you want one agent that edits well and stays out of the way.
El Crítico6.35.5Google Antigravity — Agents drive your editor, terminal and a live Chrome with no container between them and your machine; the only isolation on offer is a git worktree.
El Profesor6.35.3Google Antigravity — No benchmark is published; the design bet is verification by artifact, with plans, walkthroughs and screenshots the human reviews instead of raw tool logs.
La Inversora7.35.3Google Antigravity — Google is bundling the IDE into Google AI Pro and Ultra subscriptions, distribution no startup on this board can match and a roadmap no buyer controls.
La Jefa5.35.3no preference
El Hacker4.03.5Google Antigravity — Proprietary, local models only via a GOOGLE_GEMINI_BASE_URL override, no key field in the IDE; the agy CLI takes a Gemini API key and MCP servers sit in a mcpServers block with a serverUrl for remote ones.

Picks are derived from each critic's own scores. Humans vote on matchups on the duels page.

Want a third column? The compare tool handles any two agents. Three-way comparisons are on the roadmap once the spec rows are all verified.