The AI engine

Premium Claude. Tuned exactly how you want.

Every ovellan bot runs on Claude, from fast and efficient to our most capable flagship. Pick the model, let it route itself per question, and tune how it reasons, writes and grounds. All per bot, with a live cost readout.

Claude Haiku 4.5Standard
Claude Sonnet 5Smart
Claude Fable 5Newest flagship
Claude Opus 4.8Premium
Smart auto-routingPer question
Model choice

Choose the Claude model, per bot

Four levels, real Claude models. Start fast and efficient, step up to sharper reasoning, or run our most capable flagships on higher plans, set it per bot and change it any time.

Standard Claude Haiku 4.5

Fast and efficient, great for everyday questions.

On every plan
Smart Claude Sonnet 5

Sharp all-round reasoning for tougher questions.

Pro and up
Premium Claude Opus 4.8

Deep reasoning for complex, agentic work.

Business and up
Premium Claude Fable 5

Our newest, most capable flagship, you pick Opus or Fable.

Business and up
Smart auto-routing

Let the bot pick the model, per question

Switch it on and every question is sized up on the fly, easy ones get the fast, cheap model, hard ones get a smarter one. Best answer, lowest cost, no config. You choose which models it is allowed to route between.

“What are your opening hours?”Easy Haiku 4.5
“Compare the two plans for my use case”Medium Sonnet 5
“Draft a migration plan from my docs”Hard Opus 4.8
Answer tuning

Tune how it thinks, writes and grounds

Five dials per bot. Turn reasoning up for hard questions, keep answers short or thorough, pull in more of your knowledge, set the voice from precise to creative, and give it a human-like typing pace.

Reasoning depthFastBalancedDeepExpert
Answer lengthConciseStandardDetailedComprehensive
Context depthStandardDeepMaximum
CreativityPreciseBalancedCreative
Response paceInstantNaturalRelaxed
Enterprise self-check

A second pass that checks itself against your sources

On Enterprise, the bot re-reads its own answer against your content before sending, catching anything that isn't backed by a source. Higher confidence for high-stakes replies.

Draft answerCheck vs sourcesSend
Usage & cost

See what every setting costs, before you use it

Billing is per message. Heavier models and deeper settings use your monthly allowance faster, and a live readout shows exactly how much, up front. When your allowance runs out the bot pauses instead of overspending. No surprise bills.

HaikuSonnetOpusFable
light → uses more of your allowance
Questions

Frequently asked questions.

Which AI models does ovellan use?

Only Claude, end to end. Standard runs on Claude Haiku 4.5, Smart on Claude Sonnet 5, and Premium on our most capable flagships, Claude Opus 4.8 and Claude Fable 5. No commodity mini-model.

Do I pick the model, or does the bot?

Both are possible. Set a fixed model level per bot, or switch on smart auto-routing and let the bot pick the cheapest capable model for easy questions and a smarter one when it gets hard.

Does a smarter model or auto-routing cost extra?

Heavier models and deeper settings use your monthly message allowance faster, and you see exactly how much before you use it. Auto-routing itself adds no extra model call, it sizes the question up with a fast heuristic.

What is the self-check?

On Enterprise, the bot re-reads its own answer against your sources before sending, to catch anything unsupported. It is a second verification pass for high-stakes replies.

Put premium Claude on your site, tuned your way.

Train it on your content, pick the model, and let it answer, sell and support your customers. 7 days free, no credit card required.

AI Chatbot Models & Answer Tuning | ovellan