01 / Reasoning
Long chains without drift.
Llana sustains multi-step arguments over thousands of tokens, re-reading its own premises when it needs to.
● Llana 3.2 · Released March 2026Volume II · Issue 07
A model that thinks, patiently, in public.
Llana is Kapllan's flagship reasoning model — built for long-horizon problems where the shape of a good answer is not obvious. It reads carefully, shows its working, and prefers being right to being fast.
You
Why does ice float on water?
Llana
Most solids sink in their own liquid because freezing packs molecules closer together. Water is unusual: as it cools below 4°C, hydrogen bonds force molecules into an open hexagonal lattice that takes more room, not less. Ice is about 9% less dense than liquid water — so it floats. A small oddity, with outsized consequences: lakes freeze from the top down, and life survives the winter underneath.
§ I — Capabilities
Built for the problems that don't fit on one screen.
02 / Code
Reads codebases like prose.
128K context with structural awareness — call graphs, test intent, the difference between a bug and a choice.
03 / Research
Cites what it doesn't know.
Calibrated uncertainty — Llana will decline, hedge, or ask a clarifying question before it invents an answer.
04 / Agents
Plans. Executes. Reports back.
A tool-use interface that treats every action as revocable — Llana narrates its intent before it takes one.
05 / Vision
Describes what it sees.
Charts, diagrams, scanned pages, handwritten notes — Llana reads images with the same care it brings to text.
06 / Safety
Refuses with reasons.
Every refusal comes with a justification you can argue with — not a flat wall. Transparency is a design goal, not a patch.
§ II — Performance
On the benchmarks we trust, and the ones we don't.
| Benchmark | What it measures | Llana 3.2 | Prior SOTA |
|---|---|---|---|
| ▸MMLU-Pro | Multi-discipline reasoning | 84.1 | 81.3 |
| ▸GPQA-Diamond | Graduate science Q&A | 71.8 | 68.0 |
| ▸SWE-bench Verified | Real-world coding tasks | 62.4 | 58.9 |
| HumanEval | Code synthesis | 94.7 | 94.2 |
| ▸MATH-500 | Competition mathematics | 88.5 | 85.1 |
| AIME 2025 | Olympiad-level problems | 54.2 | 52.0 |
"We do not want a model that speaks confidently about everything. We want one that knows the shape of its own ignorance."— From the Llana 3 technical report
§ III — Research
Published openly. Replicated in the wild.
26 Mar 2026
Calibrated refusal: learning when to not answer
A. Berisha · L. Mora · H. Tanaka · et al.
Pre-print
→
18 Feb 2026
Deliberation as inference: scaling thought at test time
M. Oduya · S. Whitlock · R. Kaur
NeurIPS '26
→
07 Jan 2026
The Llana 3 technical report
Kapllan Research
Report
→
22 Nov 2025
Interpretability without the spotlight effect
J. Pell · N. Moreau · D. Orlov
ICLR '26
→
§ IV — How we build
Three working principles. All of them negotiable.
Principle 01
Slow beats showy.
We release models when their behavior is understood, not when a demo looks clean. We would rather publish a late, calibrated model than an early, charismatic one.
Principle 02
Write it down.
Every capability claim is tied to a public evaluation, a dataset, or a paper. If we cannot describe how we measured it, we do not ship it.
Principle 03
Refuse the spectacle.
Research that doesn't reduce to a screenshot is still research. A good question is a legitimate deliverable. We pay for depth.
Work with Llana.
Free during public beta. API access for researchers and developers. Enterprise pilots on request.