Orrery Nexus by Ephemerent
Agent runs,
under control.
Orrery is an agentic code editor centered on Nexus: open a workspace, pick a model route, send a task, watch the run stream, then review usage, proof, and safety receipts before you accept anything.
works with
Arbiter 27BDoublewordNexus CloudProof VaultManaged Connectors
Arbiter 27B and Doubleword credits — choose hosted routes per workspace and task.
Agent output can be wrong or unsafe; keep backups and review changes before running or shipping them.
Visible boundaryReversible workHuman acceptance
scroll ↓
01openchoose a workspace
→
02routepick a model
→
03sendrun the agent
→
04reviewproof and usage
Approval gates available · review every change · keep your own backups
Orrery Nexus
One console for workspaces, agents, routes, and receipts.
Nexus is the daily control surface: recent workspaces on the left, the active run in the center, and files, model route, agent details, proof, and usage on the right. It is built to make the common path obvious without hiding the review trail.
Illustrative lifecycle only. The sequence explains the product contract without presenting a fabricated interface or a simulated successful run.
01WorkspaceOpen a folder and initialize Nexus.
02RoutePick a hosted credit route.
03AgentChoose an eligible cloud or connected agent.
04SafetyCheck scope and permissions.
05RunWatch progress and file activity.
06ReceiptReview proof, boundary, and usage.
Daily run loop
Send work from one place, then decide what to trust.
The default flow is deliberately small: open a workspace, choose a route, send a prompt, watch the visible progress stream, then inspect summary, changed-file metadata, usage, and safety receipts. Deeper controls live in Nexus Settings.
01
Open folder
Start from the project folder you want Orrery to work in. Empty folders can be initialized as new workspaces.
02
Pick a route
Use subscription-hosted Doubleword or Arbiter routes. Arbiter 27B is the looped reasoning model, served as a hosted coding route for subscribers.
03
Run an agent
Nexus prepares workspace metadata, runs a safety check, and starts a registered agent when the route is ready.
04
Review receipts
Use proof, boundary, usage, and final summaries to decide whether to keep, revise, or discard the result.
Safety boundaries
Automation with a review boundary.
Orrery can make mistakes, and generated code may be insecure, incomplete, or unsuitable for your project. The product is designed to keep work reviewable: scope checks, permission gates, model-route labels, proof summaries, and usage receipts stay visible where supported.
Review before shippingYou are responsible for accepting, running, and deploying generated code.
No guaranteed outcomePerformance depends on task, context, models, tools, and provider availability.
Keep backupsUse source control and inspect changes before relying on them.
Cloud variesHosted routes, quotas, latency, and providers may change or pause.
Model-agnostic
Preview the setup.
Subscribe for hosted runs.
Preview setup before subscribing, then run agents through Orrery Cloud. Pro, Max, and Ultra add hosted Arbiter 27B and Doubleword credits plus managed connector automation. Pools are per plan, in the order Pro / Max / Ultra.
Arbiter 27Bsubscription hosted14M / 35M / 70M compute credits per month
Doublewordsubscription hosted6M / 15M / 30M credits per month
Nexus Cloudmanaged operationsconnectors, proof vaults, and usage receipts
Research runssubscription hostedavailable by plan and capacity
Run contractReceipt fields · stated before execution
ScopeWorkspace, route, permissions, and review boundary are named before work begins.
ActivityFile and tool events remain visible while a connected agent is working.
EvidenceTests, diagnostics, usage, and boundary notes stay distinct from generated prose.
DecisionThe run ends at human review: keep, revise, or discard.
This document describes the product contract. It is not a fabricated terminal session or a claim that a sample run succeeded.
Built for trust
Everything it does stays in view.
Orrery shows plans, tool activity, test output, review receipts, and usage data in one place. The log helps you audit the work, but it is not a substitute for your own review and backups.
01
Safety check first
Launches can record workspace scope, route state, permissions, and basic setup risk before work starts.
02
Searches your codebase
Ask about the project and Orrery can search code by meaning, not just keywords, with @codebase.
03
Scoped memory
Workspace memory is explicit and editable, so past decisions can be reused without turning every chat into hidden context.
04
Connected agents
Register local agents and connector-backed agents, then keep daily controls in Nexus and deeper setup in Settings.
05
Tool-aware runs
Use MCP and local tools where configured, while keeping raw source and keys out of cloud metadata by default.
06
Measure it over time
Track proof quality, usage, failures, and route behavior as tasks and models change.
07
Works with other editors
Plugs into other editors and agents through open, standard protocols.
08
Set the pace
Use model-aware effort and cost presets to keep routine work lean and deeper work deliberate.
09
See usage
Runs can show model route, usage, and credit receipts so hosted allotments are easier to track.
10
Cost-aware routing
Budget presets and route labels help you choose the right hosted-credit path for the job.
11
Nexus rooms
Group workspaces, agents, routes, runbooks, and approvals into reusable operating contexts.
How it compares
What a supervised run must show.
Orrery keeps the operating contract concrete: each stage has visible evidence, a bounded action, and a decision that remains with you.
| Run stage | Visible evidence | System action | Your decision |
|---|---|---|---|
| Scope | Workspace and permission boundary | Prepare the run | Approve or revise |
| Route | Provider label and credit pool | Select hosted capacity | Confirm cost posture |
| Work | Progress, tools, and changed-file activity | Attempt the task | Observe or interrupt |
| Verification | Tests, diagnostics, and proof summary | Check the result | Inspect the evidence |
| Receipt | Usage and boundary record | Close the run | Keep, revise, or discard |
The interface will keep evolving; these review boundaries are the stable product intent.
Data and account boundaries
Clear about what crosses the network.
Hosted routes send the prompt and code context needed for a completion through Orrery Cloud to the configured provider. Ephemerent meters usage, does not retain relay content beyond fulfilling the request, and does not use your code or prompts to train its own models.
01
Explicit hosted route
Nexus labels the selected Doubleword or Arbiter route before a run starts.
02
Purpose-limited relay
Only the prompt and context needed for the completion are relayed to the configured provider.
03
Content is not metering
Stored usage counters are token totals and timestamps, not copies of prompt content.
04
Your review remains final
Keep source-control backups and inspect changes before accepting, running, or shipping output.
Pricing
Subscribe to run Orrery.
Orrery is a paid product. Preview Nexus, open a workspace, and configure routes before subscribing; Pro, Max, and Ultra unlock agent runs, hosted credit pools, cloud features, managed connectors, proof vaults, and usage receipts, subject to capacity and provider limits. Hosted routes use prompt caching and token-efficient run context, so a pool of provider-cost credits goes further than the raw number suggests.
Most popular
Pro
$40
per month
Individual agent work on Arbiter 27B and Doubleword with hosted credits and Nexus operations. Zero data retention for Arbiter in one switch.
- Arbiter 27B - 14M compute credits/mo (about 12M looped tokens)
- Doubleword - 6M credits/mo (about 32M tokens at list rate)
- Credits are vendor-cost units, so nothing is marked up and nothing is inflated
- Nexus, Discord/mobile oversight, proof receipts
Max
$100
per month
Bigger pools for daily multi-agent work; the same routes, more headroom.
- Arbiter 27B - 35M compute credits/mo (about 31M looped tokens)
- Doubleword - 15M credits/mo (about 80M tokens)
- 2.5x Pro on both pools
- More cloud/background run capacity
Ultra
$200
per month
The largest pools plus cloud automation capacity for research runs.
- Arbiter 27B - 70M compute credits/mo (about 61M looped tokens)
- Doubleword - 30M credits/mo (about 160M tokens)
- 5x Pro on both pools
- Research runs and proof vault capacity
One credit is one millionth of a dollar of provider cost. Doubleword credits buy tokens at the provider's list rate; Arbiter credits buy looped inference compute (three prefill passes plus decode), because Arbiter reasons in latent space and does not bill reasoning tokens. Pools reset monthly, stop hard at the limit, and never overage. Arbiter learns from your coding and research sessions by default (anonymous session ids, never your account); switch on zero data retention in Orrery settings and it keeps counts only. See Privacy.
Organizations
Governed agent work for engineering teams.
Shared workspaces, pooled credits, Slack requests, approval gates, audit receipts, and centralized policy. An organization pool is shared across seats, reserved before dispatch and settled after, and organization entitlements sit alongside each member's personal plan.
Paid tiers are monthly subscriptions billed in advance via Stripe and renew automatically until canceled. Credits reset each billing month and do not roll over. Hosted model prompts are proxied through Orrery Cloud to provider APIs. Routes, quotas, capacity, and providers may change or be unavailable. Agent output is not guaranteed. Legacy local access is honored for existing buyers, but it is no longer sold and does not include hosted credits or cloud features. See Terms and Privacy.
Questions
Before you download.
What leaves my machine?
When you run a hosted route, the prompt and code context needed for that completion pass through Orrery Cloud to the configured provider. Usage counters store totals and timestamps; Ephemerent does not retain relay content beyond fulfilling the request or use it to train its own models. See the privacy policy for the full boundary.
Which platforms are supported?
The current desktop packet targets Windows. macOS and Linux are on the roadmap; availability and update delivery may change.
Can I use Orrery without subscribing?
You can inspect preview setup, open Nexus, prepare a workspace, and configure routes. Real agent runs require a trial, subscription, or invite grant. Legacy local access remains honored for existing buyers but is no longer sold.
Which hosted routes are included?
Pro, Max, and Ultra include monthly Arbiter 27B and Doubleword credits at the amounts shown above. Routes and capacity remain subject to provider availability.
How is the winning attempt chosen?
By execution, not impression. Each attempt is run against your tests, type-checks, and lint; an automatic reviewer scores them and keeps the winner before merging it back.
Can I cancel anytime?
Yes. Paid tiers are monthly via Stripe and cancel anytime — access runs to the end of the paid period. Token quotas reset each billing month and don't roll over.
Start in Nexus
Open a workspace.
Send one task.
Get Orrery, choose a route, and keep the run reviewable from the first prompt.