Your calibration
engineer. Offline.
Professional ECU tuning AI running locally on your PC. No cloud dependency. Your proprietary calibration data stays with you, secured by offline architecture.
Launching soon — downloads are not open yet.
See Pricing
Select Your Precision Level
Scalable local AI power for every automotive professional.
For workshop diagnostics
- memory12B parameter model
- layersPrecision and Extended profiles
- check_circleDay-to-day diagnostic reasoning
- cloud_offRuns offline after activation
- developer_boardMin 12GB VRAM (RTX 3060 12GB / RTX 4070)
- hard_drive20GB of model files
Launching soon
For professional tuners
- memory12B parameter model
- layersPrecision and Extended profiles
- analyticsHigher-precision quantisation of the same model
- cloud_offRuns offline after activation
- developer_boardMin 16GB VRAM (RTX 4060 Ti 16GB / RTX 4080)
- hard_drive24GB of model files
Launching soon
For tuning shops
- memory26–31B parameter model
- layersPrecision and Extended profiles
- settings_input_componentLarger model for complex ECU variants
- cloud_offRuns offline after activation
- developer_boardMin 24GB VRAM (RTX 3090 / RTX 4090)
- hard_drive36GB of model files
Launching soon
For R&D engineers
- memory27–31B parameter model
- layersPrecision and Extended profiles
- precision_manufacturingLargest models, highest-fidelity quantisation
- cloud_offRuns offline after activation
- developer_boardMin 32GB VRAM (RTX 5090)
- hard_drive48GB of model files
Launching soon
Engineering Precision Features
Offline Capable
Runs on your own hardware. No network round-trip and no file leaves the machine — after activation the engine answers with the connection unplugged.
Knowledge Base
2.8M+ indexed knowledge-base entries, searched online when you have a connection — included for 12 months with your license, then renewed year by year with the sync plan.
Voice Input Planned
Hands-free diagnostic queries while you work in the engine bay. Not in the current build.
Multilingual
Ask in Turkish, German or English and get the answer in the same language. Every tier runs the same open-weights model, so language coverage does not change between tiers.
Photo Analysis
Send a photo of the board or the wiring with your question — the model reads images as well as text.
Hardware ID Licensing
Secure hardware-locked activation. One license per workstation ensures maximum local stability.
Continuous Performance Updates
check_circleActive Subscription
- Model updates as new profiles ship
- Online knowledge base access
- Priority support access
- New ECU platform support
cancelNo Subscription
- Lifetime access to current model version
- Knowledge base only in the 12 months included with the license
- No community-sourced weights
- Manual diagnostic input only
| Tier Type | 3-Month Sync | 6-Month Sync | Yearly Sync |
|---|---|---|---|
| Standard | €29 | €49 | €79 |
| Pro | €35 | €59 | €99 |
| Ultra | €49 | €89 | €149 |
| Max | €69 | €119 | €199 |
Hardware Specifications
| Component | Standard | Pro | Ultra | Max |
|---|---|---|---|---|
| GPU (VRAM), minimum | 12GB | 16GB | 24GB | 32GB |
| Model files on disk | 20GB | 24GB | 36GB | 48GB |
| Operating system | Windows 10/11 | |||
| Example GPUs | RTX 3060 12GB / RTX 4070 | RTX 4060 Ti 16GB / RTX 4080 | RTX 3090 / RTX 4090 | RTX 5090 |
Each tier ships two profiles — Precision and Extended — and you pick one when you install (cl-mini mini install <profile>). The VRAM figures above are the tier's minimum requirement, set from the measured size of the model files; they are not a measurement of runtime GPU use.
Don't have powerful hardware?
Get CL-Computer.
A compact Windows 11 Pro workstation for CalibraLogic's local AI engines, with 128 GB or 192 GB of AMD unified memory.
Learn More arrow_forward
Frequently Asked Questions
Is internet connection required for daily use?
Not for chatting with the local model — that runs on your machine. A connection is needed for activation, the model download (20–48 GB, depending on tier) and a license check at least every 30 days; the knowledge base and WinOLS map identification are online features.
What is the difference between tiers?
Each tier ships its own model, sized for the hardware that tier targets — the tiers do not stack, and buying Max does not also install the Standard model. Standard and Pro both run a 12B model (Pro at a higher-precision quantisation); Ultra and Max move to 26–31B models. Larger models handle more complex ECU variants and longer multi-step reasoning; Standard is sized for day-to-day diagnostics.
Can I transfer my license to another machine?
Licenses are tied to the Hardware ID. You are permitted two transfers per year for hardware upgrades, managed through our secure portal.
Run your calibration engineer offline today.
Pick the precision level that fits your workshop.
See Pricing