NewDropstone Fast 1.622 Jun 2026
Fast continues on DeepSeek V4 Flash for 1.6. Reported as a latency-and-cost tier: its frontier-relevant benchmarks are saturated, so we make no comparative capability claim for it.
Dropstone Fast 1.6
Low-latency agentic coding for edits, refactors, and high-throughput inline completion. Built on DeepSeek V4 Flash, hosted in the US.
Built on
DeepSeek V4 Flash
Hosted in
US (SOC 2)
Approval gate
Every tool call
Refresh cadence
Monthly
Announcements
Announcements
Refresh cadenceOngoing
Every cycle we re-evaluate the open-weight leaderboard. When something better ships, Fast gets swapped. Same CLI, same security model.
Research
Overview
Everything Fast ships with.
Dropstone Fast is the tier you reach for thousands of times a day: inline completion, file-scoped edits, quick refactors, test stubs. It runs DeepSeek V4 Flash on US-hosted inference behind the same approval-gated CLI as every other tier, so velocity never costs you a security boundary. Its frontier-relevant benchmarks are saturated, so we make no comparative capability claim for Fast and report it as a latency-and-cost tier.
01
Inline completion
Low-latency suggestions that respect your repo conventions. Designed for thousand-keystroke workflows.
02
File-scoped edits
Rename, refactor, extract. Fast enough to feel like an IDE primitive, not a network round-trip.
03
Test scaffolding
Generate unit-test skeletons against the function under the cursor with the same approval gate as any other tool call.
The security boundary is the approval gate.
Every Dropstone request is treated as if the model could be adversarial. The CLI requires explicit user approval before any action that writes to disk, runs a shell command, or fetches a URL. No model output is ever auto-executed.
US-hosted inference. You control your data.
Fast runs on SOC 2-certified, US-based inference providers. Enterprise and commercial data is never stored and never trained on, regardless of any setting. Consumer sessions help improve the open model we run and can be turned off anytime in settings, as text only, never your raw images.
Honest about what we cannot prove.
Fast is built on DeepSeek V4 Flash, an open-weight foundation model. Goldwasser et al. (2022) proved no party can prove a closed foundation model is free of embedded behaviors, including Anthropic for Claude and OpenAI for GPT. We say this out loud. The runtime is why model origin does not matter for your code.
Pick your tier
Three tiers. One CLI.
Same approval gate, same US-hosted inference, same zero-retention guarantee across all three. Pick the smallest model that meets the task.
You are here
Dropstone Fast
Dropstone Fast 1.6
Low-latency agentic coding for edits, refactors, and high-throughput inline completion. Built on DeepSeek V4 Flash, hosted in the US.
Best for
- Inline completion
- File-scoped edits
- Test scaffolding
Security
- Approval gate on every tool call
- US-hosted inference
- Zero retention
- Monthly model refresh
Dropstone Pro
Dropstone Pro 1.6
The everyday workhorse, built for tool use and day-to-day coding. Built on Kimi K2.7 Code, hosted in the US.
Best for
- Full-stack feature work
- Tool-driven workflows
- Code review at scale
Security
- Approval gate on every tool call
- US-hosted inference
- Zero retention
- Monthly model refresh
Dropstone Heavy
Dropstone Heavy 1.6
The strongest open-weight model we measured: #1 open on SWE-bench Pro, ahead of GPT-5.5. Built on GLM-5.2, hosted in the US.
Best for
- Long-horizon refactors
- Hard SWE-bench-shaped issues
- Terminal-driven debugging
Security
- Approval gate on every tool call
- US-hosted inference
- Zero retention
- Monthly model refresh
Fast is included on every plan. Pro and Heavy are unlocked on the Pro plan ($15/mo) and Max plan ($75/mo).
FAQ
Questions you'd ask in a security review.
When should I use Dropstone Fast 1.6?
Dropstone Fast is the tier you reach for thousands of times a day: inline completion, file-scoped edits, quick refactors, test stubs. It runs DeepSeek V4 Flash on US-hosted inference behind the same approval-gated CLI as every other tier, so velocity never costs you a security boundary. Its frontier-relevant benchmarks are saturated, so we make no comparative capability claim for Fast and report it as a latency-and-cost tier.
Is my code sent to a Chinese model provider?
No. Inference runs on SOC 2-certified, US-based providers. The model weights are open-source and loaded into those providers' US data centers. Your prompts never touch a foreign network.
What changes when Dropstone Fast 1.6 gets refreshed?
The version number and benchmarks. The CLI surface, the pricing structure, the approval-gate behavior, and the security model do not change. Existing scripts and CI pipelines continue working without modification.
Do you offer enterprise deployment?
Yes. Dropstone Enterprise extends the same audited-tier platform with VPC, on-premises, and air-gapped deployments, SSO, audit logs, and custom SLAs. Pricing is seat plus usage at API rates, on annual commitments. Read the Enterprise plan overview or contact enterprise@blankline.org.
Ship with Dropstone Fast 1.6.
Install the CLI, authenticate, and start running approval-gated agentic workflows. No credit card to start.