Route a support ticket. Score urgency. Detect a refund request. Laya Goish turns English and Thai text into structured answers your application can use, running on your own infrastructure.
Download v0.1.0 · Get started · API reference
| Task | What you get | Put it to work |
|---|---|---|
| Choose | A choice from your options, with probabilities and confidence | Route tickets, classify messages, select a next step |
| Score | A score against your ordered criteria | Assess urgency, sentiment, or quality |
| Check | The probability that a condition is true | Detect intent, flag a request, check a requirement |
Define the questions and criteria. Send text or a JSON object. Receive structured JSON.
curl http://localhost:8080/v1/decide \
-H 'Content-Type: application/json' \
-d '{
"model": "laya",
"state": "I was charged twice. Please refund the duplicate payment.",
"questions": {
"refund": {
"type": "noul",
"instructions": "Does the customer ask for money back?"
}
}
}'
Explore example requests · Use built-in presets
Use Laya for English decisions or OpenThai-SystemOne for Thai and English. Load both into one server and select a model in each request.
| Model | Languages | Guide |
|---|---|---|
| Laya English | English | Download and run |
| OpenThai-SystemOne v0.3 | Thai and English | Convert and run |
A single Linux executable runs inference on your CPU. No hosted inference service is required. Keep the models on your machine and integrate through HTTP, the command line, or a persistent JSON-lines process.
Batch requests let you apply the same questions to a collection of inputs. Optional bearer authentication controls access to the decision API.
Download and extract the Linux release, then run:
bash scripts/download-models.sh f16
./laya serve models/laya_english_f16.gguf --port 8080 --threads 4
Requires Linux x86-64 with AVX2/FMA. Model weights are downloaded separately.
Installation and requirements · HTTP API · Release notes
Built with Goish Rust. Laya inference is ported from ggmlc; Thai/English support uses OpenThai-SystemOne.
Route a support ticket. Score urgency. Detect a refund request. Laya Goish turns English and Thai text into structured answers your application can use, running on your own infrastructure.
Download v0.1.0 · Get started · API reference
| Task | What you get | Put it to work |
|---|---|---|
| Choose | A choice from your options, with probabilities and confidence | Route tickets, classify messages, select a next step |
| Score | A score against your ordered criteria | Assess urgency, sentiment, or quality |
| Check | The probability that a condition is true | Detect intent, flag a request, check a requirement |
Define the questions and criteria. Send text or a JSON object. Receive structured JSON.
curl http://localhost:8080/v1/decide \
-H 'Content-Type: application/json' \
-d '{
"model": "laya",
"state": "I was charged twice. Please refund the duplicate payment.",
"questions": {
"refund": {
"type": "noul",
"instructions": "Does the customer ask for money back?"
}
}
}'
Explore example requests · Use built-in presets
Use Laya for English decisions or OpenThai-SystemOne for Thai and English. Load both into one server and select a model in each request.
| Model | Languages | Guide |
|---|---|---|
| Laya English | English | Download and run |
| OpenThai-SystemOne v0.3 | Thai and English | Convert and run |
A single Linux executable runs inference on your CPU. No hosted inference service is required. Keep the models on your machine and integrate through HTTP, the command line, or a persistent JSON-lines process.
Batch requests let you apply the same questions to a collection of inputs. Optional bearer authentication controls access to the decision API.
Download and extract the Linux release, then run:
bash scripts/download-models.sh f16
./laya serve models/laya_english_f16.gguf --port 8080 --threads 4
Requires Linux x86-64 with AVX2/FMA. Model weights are downloaded separately.
Installation and requirements · HTTP API · Release notes
Built with Goish Rust. Laya inference is ported from ggmlc; Thai/English support uses OpenThai-SystemOne.