Cherry is the self-hosted alternative to Msty.
Msty is a desktop AI system: Studio for chat + knowledge, Go for scoped agents, Nexus for a local model gateway with credential control and a Stack knowledge layer on the roadmap. It's model-agnostic and privacy-focused and ships to individuals and teams via desktop install with SSO and audit logs at the Nexus / Enterprise tier. Strong story for local-model workflows and privacy-first single-machine work. Cherry is the cloud-native counterpart: a self-hosted web workspace with real multi-user teams, browser access from anywhere, per-message UsageEvent attribution, private MinIO file storage with presigned URLs, the Pins/Clipboard/Snippets three-tier reuse system and modular Apps. Deploy on your hardware, run it as your own SaaS, do whatever you want. You own the whole stack.
Msty ships as a downloaded app per machine. Cherry runs in any browser.
Cherry deploys on your hardware, in your cloud or as your own SaaS.
Postgres entities power the Cherry workspace.
Four places a desktop-first workspace runs out of runway.
Msty nails the privacy-first desktop story with a strong local-model gateway. The trade-off is that a desktop-first architecture makes team collaboration, browser-anywhere access and centralized ops harder than they need to be.
Desktop per machine
Msty ships as an installer per operating system: every user, every laptop, every reinstall. Cherry is web-first, deployed once on your infrastructure and reached from any browser. It's the same workspace on your desktop, laptop, iPad or a client's borrowed screen. There's no install per user and no version drift between machines.
Local sync, not shared workspace
Msty's default is a private per-machine workspace with sync between your own devices. Real team collaboration needs the Nexus or Enterprise tier. Cherry ships multi-user, multi-workspace with Owner/Admin/Member roles from day one. Every user hits the same Postgres, sees the same folders, shares the same Snippets.
Local knowledge, not team knowledge
Msty's Knowledge Stack lives on the desktop it was ingested from. Cherry's file surface is MinIO on your infrastructure with presigned URLs, sanitized filenames and bulk folder upload that replicates an entire OS tree transactionally. Every workspace member sees the same corpus, addressable from every chat.
Nexus for control, not attribution
Msty's Nexus gateway centralizes model credentials and policies. That's control. Cherry adds full attribution on top: a UsageEvent per message with tokens, dollar cost and latency attributed to user, workspace, chat and model. Consultants attribute AI spend per client engagement in a single query, not just "who has permission".
The web workspace Msty's desktop story deliberately isn't.
Msty nails privacy-first desktop workflows with strong local-model support through Studio, Go and the Nexus gateway. Cherry answers a different question: what if the whole team needs the same workspace from any browser on any device? One Postgres database instead of per-machine sync. One MinIO store instead of per-desktop Knowledge Stack. Multi-workspace with roles from day one instead of Nexus-tier collaboration. UsageEvent attribution instead of just credential control. And a Docker Compose deploy that runs on your hardware, in your cloud or as your own SaaS.
Msty vs Cherry.
A desktop-first privacy workspace versus a cloud-native self-hosted workspace built for real teams.
| Feature | Msty | Cherry |
|---|---|---|
| Delivery | Desktop app (Studio + Go + Nexus) | Web workspace, deployed via Docker Compose |
| Access | Per-machine install | Any browser, any device |
| Model providers | Model-agnostic (local + hosted) | OpenAI + Anthropic shipping, typed adapter for any provider |
| Local models | First-class (Ollama etc) | Pluggable via provider adapter (Ollama roadmap) |
| Deployment | Desktop install per user + Enterprise gateway | Self-hosted. On-prem, cloud, run as your own SaaS. You decide |
| Storage | Local disk with sync | PostgreSQL + MinIO on your infrastructure |
| Data model | Workspaces + prompts + Knowledge Stack | 15+ entity Postgres schema (Workspaces, Presets, UsageEvents, more) |
| Team surface | Nexus tier + SSO + audit logs | Multi-workspace with Owner/Admin/Member roles from day one |
| Folder tree | Workspaces + folders | Infinite nested self-referential parents |
| Per-message bookmarks | No | Pins with scroll-to-highlight animation |
| Per-chat clipboard | No | Drag-reorderable saved passages, promote to Snippet |
| Cross-chat snippets | Prompts library (local) | Tagged, folder-scoped, one-click injection, source-tracked |
| Split chat / Compare | Split chat across models | Compare modal fires one prompt at every provider side by side |
| Knowledge / RAG | Knowledge Stack (versioned, roadmap) | Files + Snippets addressable per chat, RAG roadmap |
| Agents | Go (scoped, step-by-step review) | Agent Library with Presets across six categories |
| Credential gateway | Nexus (centralized key + policy) | Server-side keys with per-workspace override at Enterprise |
| Cost per prompt | Not first-class | Tokens, dollar cost, latency on every message |
| Usage dashboard | Basic | Recharts. Today / 7d / 30d / billing cycle |
| Bulk folder upload | Per-desktop ingest | Replicates an entire OS folder tree transactionally |
| SSO | Nexus / Enterprise tier | Available in Enterprise, configurable on your infrastructure |
| Ownership | Freemium + Enterprise licensing | You own the code, the data, the deployment |
Frequently asked questions.
Does Cherry run local models like Msty does?
Not out of the box today. Cherry ships OpenAI + Anthropic adapters and hits hosted APIs directly. Because the provider abstraction is a typed interface, wiring an Ollama or vLLM adapter is a single-file addition and is on the roadmap. If local-model-first is the primary requirement, Msty is a strong fit today; if team-first + browser-accessible + real data model matters more, Cherry is the answer.
How does Cherry replace the Nexus gateway?
Cherry ships server-side keys shared across a workspace by default. Enterprise deployments support per-user or per-workspace BYO keys with the same central-control story Nexus offers. Cherry adds full UsageEvent attribution on top, not just "who has permission" but "who spent what where".
What replaces Msty's Knowledge Stack?
Cherry ships files stored in MinIO with presigned URLs and bulk folder upload for ingesting an entire OS tree at once. Files are addressable from any chat by any workspace member. RAG over the file corpus is on the roadmap; the file surface itself ships today.
Can we run Cherry on our own hardware for privacy?
Yes. That's the default deployment model. Cherry ships as Docker Compose. It orchestrates nginx, the Next.js frontend, the Express backend, Postgres, Redis and MinIO. Deploy on your own hardware, in your cloud or as your own SaaS product. Data never leaves your infrastructure.
Does Cherry ship an agent system like Msty Go?
Cherry's Agent Library is a Preset system with model + temperature + system prompt + tier requirement across six categories. Step-by-step review-style agents with tool use are on the roadmap. The current shipping surface is Presets plus modular Apps (Canvas, Code Interpreter beta, Image Generation).
What does Cherry cost?
Contact sales for pricing. Every engagement is scoped to how you plan to deploy, who your users are, which provider adapters you need lit up and what support terms fit your organization.
Ready to work in the same workspace from any browser, on any device?
Tell us how your team uses Msty today, which local models you rely on and where the desktop-first model starts costing you. We'll respond within one business day with an honest read on whether Cherry fits.