Behavioral Routing
One local model, three behaviors. Rada adjusts the system prompt, temperature, and context window based on whether you're refactoring, building, or learning.
Local-first AI coding · Now accepting early access
Your AI. Your machine.
Cloud when you need it.
A desktop AI coding workspace that runs local models by default and routes to cloud only when the task demands it. One model in RAM, zero model swapping.
Cloud AI coding tools keep raising prices and gating usage. Rada gives you local-first AI coding without the cloud lock-in.
Native app for macOS, Windows, and Linux · No spam · Unsubscribe any time
EK
JS
MR
The problem
Every AI coding tool is cloud-only. The cost model is breaking.
You're paying cloud prices for refactors, explanations, and quick fixes that could run on your own hardware. And when providers can't sustain it, they raise prices or cut access.
# What cloud-only AI coding looks like # refactor_function() → cloud API → $0.03 # explain_this_code() → cloud API → $0.02 # fix_typo() → cloud API → $0.01 # What Rada does instead rada.route("local") # your machine, your model, $0
What Rada does
Smart cloud routing
When a task exceeds local capability, the Autorouter picks the right cloud endpoint automatically. Routed requests burn at half the normal quota rate.
Persistent memory
Context, decisions, and code patterns survive session restarts. Your AI assistant actually knows your codebase without being reminded every time.
How Rada compares
Capability
Cursor / Claude Code / Codex
Rada
Where models run
Cloud only
Local-first, cloud when needed
Model switching
Hot-swap (RAM spikes)
One model, behavioral routing
Cloud cost control
Monthly subscription, usage-gated
Daily burst quota, 0.5x autoroute rate
Session memory
Resets on close
Persists across sessions
Pricing
Local is always free. Paid plans add managed cloud routing with a daily burst quota. All prices in USD.
Free
$0USD · forever
No card required
Local models on your own machine.
- Local models only — runs entirely on your hardware
- Bring-your-own-key cloud routing (your API keys, your bill)
- Persistent session memory
- No Daily Cloud Burst
Included with the app
Get early access →Pro
$19USD / month
or $149 USD / year billed annually
Managed cloud routing when a task outgrows local.
- Everything in Free
- Managed cloud model access — no API keys needed
- 20 Daily Cloud Burst (40 effective with Autorouter)
- Burst Day uplift on the first Tuesday of each month
Available in the app
Join the waitlist for access →Ultra
$55USD / month
Billed monthly
Priority routing and heavyweight models.
- Everything in Pro
- 75 Daily Cloud Burst (150 effective with Autorouter)
- Priority routing queue
- Heavyweight cloud models
Available in the app
Join the waitlist for access →Prices are in USD and exclusive of VAT and sales tax, which is added at checkout where applicable. Payments are processed by Creem.io, the merchant of record for all Rada purchases.
Monthly and annual plans renew automatically until cancelled. Refunds are available within 14 days of the original charge — see the refund terms in our Terms of Service.
Checkout happens inside the Rada desktop app. The app is in closed beta — join the waitlist below for access.
Beta perks
For closed-beta members
Beta testers go first for Ultra Lifetime cloud access.
Beta testers help us shape Rada, so beta testers are first in line when the one-time Ultra Lifetime tier opens for sale. Lifetime is one payment for cloud access at the Ultra level — forever — and as a beta participant you'll be invited before it goes public. Lifetime pricing will be published here when it becomes available to buy.
- Early-access invite to the Lifetime tier when it goes live
- Direct line to the team during the beta — your bugs, your features
- Founding-supporter recognition once we exit beta
No price commitment to join the waitlist — the beta is free to try, and paid plans are only ever opt-in.Join the closed beta →
Local-first AI coding.
No cloud lock-in.
Closed beta coming soon. Get early access.