01
Born Writer
RESEARCHA compact, local story engine for agentic roleplay, character, continuity, and genuinely human prose.
Qwen3.6 35B-A3B / multi-teacher foundry / SFT + preference + world-feedback RL
Read the technical planAI made for Irish reality.
Open models, evaluation, Gaeilge research and private fine-tuning from one independent lab in Waterford.
LeemerLabs systemActive
MODEL LAB
Born / open model adaptation
MEASUREMENT
BornBench / release gates
DEPLOYMENT
European inference / private endpoints
LANGUAGE
Gaeilge data / models / evaluation
The lab
One lab. Working programmes.
Born is now the applied model programme inside LeemerLabs. The infrastructure, research and commercial work share one system.
02
Born models
LIVEOur applied model programme. Train, measure, release, document, repeat.
Born-9B Preview / public adapter / 23 of 25 local gate passes
Explore Born03
Evaluation
OPENBornBench, held-out gates and task-specific harnesses that keep claims tied to evidence.
Methods, answer sheets and scored result files published
Open BornBench04
Experimental chat
OPENThe public workbench for trying Born lanes and comparison models against real prompts.
Model switching, local or cloud history, usage windows and feedback-ready conversations
Open experimental chat05
Inference
BETAOpenAI-compatible access to open-weight models through a European data path.
API keys, usage dashboard and developer-ready endpoints
Request access06
Open releases
PUBLICAdapters, model cards, provenance notes and evaluation artifacts people can inspect.
Hugging Face releases backed by technical write-ups
Read the workFine-tuning service
Your data. Your weights. Evidence included.
We adapt open models for a defined task, then deliver the model, training record, evaluation and deployment path.
Fine-tuning servicesFrame the task
Define the behaviour, constraints and evidence that would make a custom model useful.
Build the data
Create, clean, license and split the examples that carry the domain into training.
Adapt the model
Run LoRA, QLoRA, distillation or full tuning against an appropriate open base.
Measure and deliver
Compare against the baseline, document failures and package deployable weights.
Gaeilge AI programme
A language should not need permission to enter the future.
We plan to build with fluent speakers, researchers and institutions, publishing what can be open and protecting what must remain private.
Open the Gaeilge programmeLISTEN
Work with speakers and language organisations before setting model objectives.
CURATE
Build licensed, documented datasets and evaluation sets for real Irish usage.
ADAPT
Fine-tune strong open multilingual bases for Irish and bilingual tasks.
MEASURE
Publish baselines, failures, model cards and reproducible evaluation methods.
RELEASE
Open the artifacts that can strengthen the wider Irish-language ecosystem.
Writing
Work in public.
Gaeilge / 12 Jul 2026
The future of the Irish language in the AI era
Irish needs more than a chatbot demo. It needs licensed data, fluent-speaker evaluation, open models, durable institutions, and tools people choose to use.
Read articleOpen source / 12 Jul 2026
Open source is language infrastructure
For a lower-resource language, inspectable weights, documented datasets, open evaluations, and community governance are practical safeguards against digital dependence.
Read articleLaunch / 17 May 2026
Born-9B Preview is finally here
The launch note for Born-9B Preview: Repath at the controls, GPT-5.5 Codex managing the run, the datasets, failures, benchmarks, and the public adapter.
Read article