CI/CD and code review
Builds, tests and an AI review on every pull request, on runners in your office. Only the review comment goes back to GitHub.
$ git push origin fix/flaky-retry build ✓ tests ✓ review ✓ cost $0.00
Lucebox for business.
Local AI for the whole team, on machines you own. We install them on your network, tune the models you rely on, and your engineers talk to ours directly. Every Business deal starts with a call.
Two ways to buy.
| On your own Developer, for individuals | With our support Business, for teams and companies | |
|---|---|---|
| Starting at $5,999$6,899€6,749£5,999 USD per machineUSD per machineEUR per machineGBP per machine Taxes, duties and shipping includedShipping and duties includedTaxes, duties and shipping includedTaxes, duties and shipping included | Price Custom Volume pricing for your fleet Set on a call | |
| Order now Card or Shop Pay Full refund within 30 days Terms | Book a call Purchase order and invoice Written quote and contract | |
| Pricing | ||
| Machine, 128 GB and 2 TB | $5,999$6,899€6,749£5,999 | Volume pricing |
| 192 GB memory | +$2,200+$2,600+€2,450+£2,000 | Volume pricing |
| 4 TB storage | +$250+€279+£250 | Volume pricing |
| Cluster kit, per machine | +$799+€869+£799 | Volume pricing |
| Written quote and contract | ||
| How you pay | Card or Shop Pay | Purchase order and invoice |
| Setup | ||
| Setup from a browser in about ten minutes | ||
| Installed on your network by us | ||
| Cluster kit fitted before shipping | ||
| Cluster set up on your network by us | ||
| Models | ||
| Tuned models preinstalled | ||
| Your own models and fine-tunes, tuned by us | ||
| Support | ||
| Email support | ||
| Discord community | ||
| Private chat with our engineers | ||
| Support response time in the contract | ||
| Engineering time for your integration | ||
| Security review for your environment | ||
| Hardware | ||
| 72-hour test before shipping | ||
| 1-year warranty, parts and labor |
What we do for your team.
| Install | We set the machines up on your network, with accounts, models and clients ready at handover. |
|---|---|
| Your models | Your models and fine-tunes, tuned and packaged for the machine. |
| Integration | Engineering time for your agents, RAG pipelines, CI and internal tools. |
What teams run on it.
Coding agents
Hand agents the long jobs, like a refactor or a test suite, and run several at once with no rate limit. One click in Lucebox Manage opens Claude Code, Codex or OpenCode already pointed at the box.
connect claude code → lucebox.localmodel qwen3.8-27b> write tests for the upload retry logic
Private knowledge base
Ask questions across your wiki, specs, tickets and code, and get answers that cite the file. Contracts, patient records and lab data can be searched the same way, because nothing leaves the box.
> where do we set the upload retry limit? maxRetries = 5, services/upload/config.ts source: docs/runbooks/uploads.md
Always-on agents
Agents like Hermes or OpenClaw triage new issues, watch the alerts and answer in Slack or Discord while the team is offline, then leave a summary for the morning. Leaving them on all night costs only power.
07:00 overnight summary 9 issues triaged, 2 marked urgent disk on db-2 at 91%, ticket opened
What our CI costs.
We priced our builds, tests and AI code review on GitHub's runners and hosted DeepSeek V4.1 Flash, then on two Luceboxes. Move the slider to your own monthly bill.
- Pays for itself in
- about a year
- Then saves
- $1,170 a month
- Over 36 months
- 2.7× cheaper
How we worked it out
- The slider starts at our own CI, 12 days of logs scaled to a month: about $1,165 of build and test minutes at GitHub's list prices and $65 of AI code review on hosted DeepSeek V4.1 Flash.
- GitHub's runner prices, checked 20 October 2026: $0.006 a minute for 2-core Linux, $0.010 for 2-core Windows, $0.026 for 16-core ARM and $0.052 for GPU.
- DeepSeek V4.1 Flash on OpenRouter at its most common price: $0.30 per million tokens read, $0.006 cached and $1.20 written. About two thirds of what our reviews read is cached.
- A 128 GB box runs the builds ($5,999) and a 192 GB box runs DeepSeek V4.1 Flash for the reviews ($8,199). One build box carries about 1.8 times our builds and one AI box about 3.4 times our reviews, so a bigger bill adds boxes. Power is about $2 a day for both, at $0.30 per kWh.
- Most of the saving is in the builds. At our volume hosted DeepSeek V4.1 Flash costs less than the AI box; the box keeps the code in the office. US prices. Our full post on this CI is coming soon.
Questions teams ask.
How much does Business cost?
It depends on the number of machines, the configuration and the support you need. We agree the price on a call and put it in a written proposal, with one price for the whole fleet.
Which models can we run?
Every model the Developer machine runs, 5 families tuned today, plus your own models and fine-tunes, which we tune and package for the machine.
Does any data leave our network?
No. The models run on your machines, and the API answers only on your network and your tailnet.
How many people can one machine serve?
With five requests at once each still gets about 60 tok/s, enough for about 10 agents on one box. More machines, or a cluster, take more.
How do we pay?
Purchase orders and invoices for your procurement team, on terms agreed in writing.
How it starts.
-
A call
Tell us about the team, the models and the work. We tell you what fits.
-
A proposal
Machines, configuration, models to tune, support terms and price, in writing.
-
Install and support
We ship, install and hand over, then stay on chat with your team.