Documentation v2.3 · Updated May 2026
Build private AI. Own everything.
The complete reference for fine-tuning, deploying, and scaling domain-specific AI models — without giving your data to anyone.
100+
Base Models
Llama · Mistral · Qwen · more
< 38 min
First Model Trained
LoRA on A10G, 20K pairs
99.9%
API Uptime
SLA-backed inference
100%
Data Private
Zero egress · Air-gap ready
Get running fast
From data to deployed model in 4 steps.
terminal — step 01
Install the SDK
$ pip install langtrain
Collecting langtrain...
Successfully installed langtrain-2.3.0
$ langtrain --version
Langtrain v2.3.0 · Python 3.11 · CUDA 12.1
All data stays in your environment · Zero egress
Popular guides
Where most engineers start.
Integration options
Connect from any language or tool.
pip install langtrain
Python SDK
from langtrain import Langtrain
client = Langtrain(api_key='lt_...')
resp = client.chat(
model="cs-agent-v1",
messages=[{"role": "user",
"content": "Hello"}]
)
print(resp.choices[0].message.content)
Full-featured SDK for training, inference, and monitoring.Docs
All documentation
Browse by topic.
Training
LoRA, QLoRA, full fine-tune — pick your method and GPU budget.
Deployment
Push to AWS, GCP, Azure private VPC or your own hardware.
SDKs & APIs
Python, Node.js, and a fully OpenAI-compatible REST interface.
Agents
Build RAG pipelines, tool-calling agents, and multi-step workflows.
Monitoring
Track latency, hallucination rate, and model drift in production.
Security
HIPAA, SOC 2, GDPR — compliance built in, not bolted on.
Free to start · No credit card required
Ready to train your first model?
Your model. Your data. Your weights. Running in under 38 minutes.