Skip to content

FAQ

In total 24 answers. Can’t find yours? Email us — download, installation and licensing questions all get replies. Full pricing rules live on the pricing page, and quotes for human services on the services page.

Pricing and membership

What's missing from the free tier?
Nothing. Modeling, all 5 analysis types, post-processing, reports, all three code packages and script export are fully enabled. The only limit is ≤ 5,000 elements per model.
What can 5,000 elements actually handle?
Enough for portal frames, 3-storey frames, small space trusses and 5-storey office buildings. Above 8 storeys with a fine mesh, and fine shell models, you need membership. See the full conversion table on the pricing page.
What happens over the limit? Will I lose my model?
No. Files save and open as usual, existing results stay viewable, reports still export — you simply can't solve that over-limit model. You can subscribe, or coarsen the mesh.
Does membership include AI?
No. AI runs on your own keys or local models — local models cost nothing at all. We never resell model fees, and we never touch your keys.
Will I be charged automatically?
Not by default. Charging happens only if you explicitly opt into auto-renewal, and you'll be reminded before expiry (exact days on the pricing page). Turn it off anytime.
Can I work offline?
Yes — license caching lasts 30 days. Intranets, construction sites and firewalled environments all follow this rule; we never require a connection for each solve.
How many machines can I install it on? What if I switch computers?
3 machines. Unbind and rebind yourself from the account page — takes effect instantly, with no limit on how often.
Will prices go up in the future?
Possibly, but with 30 days' advance notice, and within an already-paid period you're unaffected. The free tier's capability boundaries are written in the table on the pricing page and will not shrink.
Why tier by element count instead of by features?
Feature tiers force you to pay for capabilities you never use on small models — and give us an incentive to hide features. Limiting only scale means everything works if your model fits — which is exactly why we can say the free tier is feature-complete.

Installation and running

Antivirus blocks or deletes the installer
Usually heuristic blocking triggered by an unsigned installer. Add the installer to your trust list, or verify the SHA256 against the download page before allowing it to run.
Missing runtime library message after double-clicking
The installer bundles the VC++ runtime; on very old systems, install the Microsoft Visual C++ 2015–2022 Redistributable (x64) first.
Launch fails when the install path contains Chinese characters or spaces
Install to the default user directory. If you choose a custom path, avoid Chinese characters and special symbols — the OpenSees subprocess is path-sensitive.
The login verification code never arrives
Check spam first; corporate mail servers may filter it. Try a mainstream mailbox domain. Codes are valid for 10 minutes and can be resent.
The auto-update download is stuck
Click 'Retry' in settings; if it still fails, download the latest installer manually and install over the top — project files and your license are unaffected.
The solver reports convergence failure — what now?
First check the red-flagged items from the model checker (orphan nodes, unrestrained DOFs, zero-length elements, etc.) — most non-convergence is a model problem, not an algorithm problem. Then review the step size and iteration limits in the solver controls. A convergence troubleshooting doc is on the way; for now, feel free to e-mail us.
Do I have to install OpenSees myself?
No separate installation needed. The installer bundles the current kernel, OpenSees 3.7.2, and all required runtime libraries — ready to run out of the box.

Data and privacy

Does the AI upload my model?
With local models (Ollama / LM Studio / vLLM) data never leaves your machine, at any step. With cloud APIs, what goes out is the model summary and conversation you asked it to read — not the whole project file. The exact scope depends on your chosen provider; keys and billing stay in your hands.
Where is my API key stored?
Encrypted in the OS keychain (Windows DPAPI / macOS Keychain). Never uploaded, never logged, never stored in .lintel project files — sending a project to someone never carries your key with it.
What does the update check send?
Only the version number and operating system — no project information, no files.
Why is an e-mail required for registration? Can't I use a phone number?
E-mail only, no phone numbers, no real-name verification. Login carries subscription state, device binding and order ownership — none of which work without an account. Registration happens at first launch, not before download.
Do my model files survive uninstalling?
Yes. .lintel project files and .stres result files are ordinary files; the uninstaller never deletes your data.

Documentation and support

Why is the AI's knowledge base public?
Because 'verifiable results' is a baseline promise of this product — computed results must survive order-of-magnitude checks, and AI statements must have checkable sources. Hiding the AI's knowledge sources amounts to asking you to trust an unsourced conclusion.
What if the documentation and the software disagree?
The AI knowledge base and the desktop app read the same files, so in theory they cannot disagree. If it ever happens, trust the software and e-mail us which entry — this is the class of conflict we most want to fix.
Does membership include free technical support?
No — the two are priced separately. Membership is a software license (lifting the element limit); support is people's time. See /services for consulting rates; research-group members get it free.

Still stuck? Email support@kavvo.app For download, installation and licensing issues, email us directly with your version number and a screenshot. with your version number and a screenshot. Curious how large a model Kavvo can solve — see the size conversion table.