Free web
For: Trying the agent
Limits
- Limited monthly usage
- Fast and Pro models with XHigh Thinking
Includes
- CLI and Dashboard
- Web search in agent
2 more
- Memory across conversations
- Remote MCP
CodingDocumented overviewLast checked Oct 1, 2026
AI coding agent by Blankline with persistent memory across CLI, chat, SDK and editor, running on hosted open-weight models.
(open full size in a new tab)
(open full size in a new tab)
(open full size in a new tab)
(open full size in a new tab)
(open full size in a new tab)
(open full size in a new tab)Our take: Choose Dropstone if you want a terminal-first coding agent with one memory shared across CLI, chat and editor extensions, starting free and moving to Pro at $15 per month for the Heavy model tier.
Main limitation: Free usage is limited monthly, and consumer plans have session-content programs for training Dropstone's own models switched on by default.
Choose Dropstone if you want a terminal-first coding agent with one memory shared across CLI, chat and editor extensions, starting free and moving to Pro at $15 per month for the Heavy model tier. Based on the vendor's documentation, not on a hands-on test.
Also a good fit for
Or if
Prices checked Oct 1, 2026 on the official US pricing page. Official pricing ↗ Prices and limits change often.

Free plan with CLI and Dashboard and limited monthly usage.
No trial stated; Free plan is the entry.
For: Trying the agent
Limits
Includes
$15/mo source (opens in a new tab)
Month
Billed monthly; save 17% annually
For: Individual developers
Limits
Includes
$75/mo source (opens in a new tab)
Month
From $75/mo; Max 5x or Max 10x; billed monthly
For: Heavy daily users
Limits
Includes
Custom pricing source (opens in a new tab)
For: Teams and organizations
Limits
Includes
Everything is in-tier with no add-ons; Max price above $75 depends on 5x vs 10x tier (only the from price was shown).
Not stated.
Dropstone is a coding agent from the research lab Blankline that keeps one memory per account across its CLI, web chat, SDK and VS Code/Cursor/Windsurf extension. It offers Fast, Pro and Heavy model tiers built on open-weight models, a 1M-token context window on every tier, plan mode, sandboxed code execution and approval gates on tool calls. Beyond code, the agent can watch an inbox, control smart-home devices and place phone calls (Max plan). Managed inference is US-hosted with zero data retention. The CLI installs as a single binary on macOS, Windows and Linux; a Free plan has limited monthly usage, and Pro/Max plans raise limits and add the Heavy tier and priority routing.

You provide
You get back
Scenarios are worked out by our editors from the documentation; the actual result depends on your input and plan.

Stores derived knowledge, lessons and summaries against your account and reloads them in CLI, chat, SDK and extension sessions. No re-teaching
Routes work to Fast, Pro or Heavy models, all running on hosted open-weight models with XHigh Thinking effort. Trade speed for depth
Loads up to one million tokens of repository and conversation into each request. Holds large repos
Runs code in a sandbox and pauses for your approval on each tool call, with plan mode to preview changes first. Safe edits

Find why the checkout total is off by one cent, propose a fix in plan mode, then apply it and run the unit tests.

Dropstone has no public API in its official documentation. Docs list 31 CLI reference articles including the API API documentation ↗
Summarized from the vendor's current policies; this is not legal advice and a vendor's badge is not an independent audit.
Only relevant code snippets are sent to inference endpoints; indexing and embeddings run locally.
Policy says no third-party training and stateless inference, but Free, Pro and Max accounts are opted by default into programs using outcome metadata and full coding sessions to post-train Dropstone's own models; can be turned off in dashboard settings; Enterprise excluded.
Inference is ephemeral; derived memory (knowledge, lessons, summaries) is retained against your account.
Enterprise excluded from training programs.
Policies: dropstone.io/privacy
"Better suited" is based on documented features and pricing, not on a head-to-head test.
| Alternative | When to choose it | Pricing | Limitation | |
|---|---|---|---|---|
| Cursor | — | — | Compare | |
| GitHub Copilot | — | — | Compare | |
| Devin Desktop (formerly Windsurf) | — | — | Compare | |
| Claude | — | — |
No user reviews yet.
Used Dropstone? Be the first to share your experience.
They are three open-weight model tiers; Free includes Fast and Pro with XHigh Thinking and Heavy needs the Pro plan or higher. source (opens in a new tab)
The pricing page says 6x more usage than Free in the card and 8x Free in the comparison table; Max is 5x or 10x Pro. source (opens in a new tab)
Third-party model providers do not train on it, but Free, Pro and Max accounts participate by default in programs where session content may post-train Dropstone's own model; you can turn them off in dashboard settings. source (opens in a new tab)
A VS Code, Cursor and Windsurf extension exists alongside the CLI, chat and dashboard, all sharing one account memory. source (opens in a new tab)
Real phone calls on your behalf are on Max; smart home control starts at Pro. source (opens in a new tab)
Not verified: Max 10x price; Exact usage numbers; Supported languages.
Every fact about Dropstone above comes from the 4 official pages listed here, last compared on Oct 1, 2026; we have not run Dropstone ourselves, so it has no editorial score. How we check and re-check pages.
Article updated Oct 10, 2026; prices checked Oct 1, 2026. Outbound links to vendors are plain links; we do not earn commissions. How we review · Report an error
compare>