laya

10 TagsUpdated Sep 23, 2026Apache-2.0by Convai Innovations

Open decision models from Convai Innovations. Typed, calibrated answers to choice, score and yes/no questions in a single forward pass, in English and 100+ languages.

multilingualrouterguardrailsfine-tuned322m421m

Pre-release. Ollaya's first release is in progress; these commands will work once it ships. Star / watch on GitHub to hear about the first release.

ollaya run laya --preset triage "I was charged twice for my subscription this month and want a refund."

curl http://localhost:11435/api/decide \
  -H "Content-Type: application/json" \
  -d '{
    "model": "laya",
    "state": "I was charged twice for my subscription this month and want a refund.",
    "questions": {
      "department": {
        "type": "choice",
        "instructions": "Which team should handle this?",
        "criteria": {
          "billing": "Payments, invoices and refunds",
          "technical": "Bugs, errors and outages",
          "account": "Login, profile and settings"
        }
      },
      "refund": {
        "type": "noul",
        "instructions": "Is the customer asking for a refund?"
      }
    }
  }'

# Already using a TypeSafe SDK? Set TYPESAFE_BASE_URL=http://localhost:11435 instead.
import requests

response = requests.post(
    "http://localhost:11435/api/decide",
    json={
        "model": "laya",
        "state": "I was charged twice for my subscription this month and want a refund.",
        "questions": {
            "department": {
                "type": "choice",
                "instructions": "Which team should handle this?",
                "criteria": {
                    "billing": "Payments, invoices and refunds",
                    "technical": "Bugs, errors and outages",
                    "account": "Login, profile and settings"
                }
            },
            "refund": {
                "type": "noul",
                "instructions": "Is the customer asking for a refund?"
            }
        }
    },
)
answers = response.json()["answers"]
print(answers["department"]["choice"], answers["refund"]["noul"])

const response = await fetch("http://localhost:11435/api/decide", {
  method: "POST",
  headers: { "Content-Type": "application/json" },
  body: JSON.stringify({
    model: "laya",
    state: "I was charged twice for my subscription this month and want a refund.",
    questions: {
      department: {
        type: "choice",
        instructions: "Which team should handle this?",
        criteria: {
          billing: "Payments, invoices and refunds",
          technical: "Bugs, errors and outages",
          account: "Login, profile and settings"
        }
      },
      refund: {
        type: "noul",
        instructions: "Is the customer asking for a refund?"
      }
    }
  }),
});
const { answers } = await response.json();
console.log(answers.department.choice, answers.refund.noul);

Models

View all

Name	Size	Context	Input
layalatestrouter	—	512 / 1024	Text · auto (en / multilingual)
laya:en~0.8 / 1.7 GB · 512 ctx · English	~0.8 / 1.7 GB	512	Text · English
laya:multilingual~0.65 / 1.3 GB · 1024 ctx · 100+ languages	~0.65 / 1.3 GB	1024	Text · 100+ languages
laya:typed-decisions~0.8 / 1.7 GB · 1024 ctx · English	~0.8 / 1.7 GB	1024	Text · English

Sizes are approximate (fp16 / fp32). Bare tags pull fp16 on a GPU and fp32 on CPU.

Readme

Laya is a family of open decision models by Convai Innovations, released under the Apache-2.0 license. A decision model reads a state — a message, an email, a support ticket, a JSON object — together with a set of typed questions, and returns a typed answer with calibrated probabilities for every question in a single forward pass. It never generates text.

Models

Tag	Backbone	Params	Context	Languages	Best for
`laya:latest`	router	—	512 / 1024	auto	Picks `en` or `multilingual` from the detected script and language
`laya:en`	ModernBERT-large	421M	512	English	Guardrails, email triage
`laya:multilingual`	mmBERT-base	322M	1024	100+	Non-English and mixed-language input; up to ~2.2× faster on batched calls
`laya:typed-decisions`	ModernBERT-large	421M	1024	English	Typed-decisions workflows (fine-tuned)

Tags without a suffix resolve to fp16 on a GPU and fp32 on CPU. Add -fp16 or -fp32 to choose explicitly, for example laya:en-fp32.

Usage

ollaya run laya --preset triage "I was charged twice for my subscription this month."

Or call the local API:

curl http://localhost:11435/api/decide \
  -H "Content-Type: application/json" \
  -d '{
    "model": "laya",
    "state": "I was charged twice for my subscription this month.",
    "questions": {
      "department": {
        "type": "choice",
        "instructions": "Which team should handle this?",
        "criteria": {
          "billing": "Payments, invoices and refunds",
          "technical": "Bugs, errors and outages",
          "account": "Login, profile and settings"
        }
      },
      "urgency": {
        "type": "score",
        "instructions": "How urgent is this?",
        "criteria": ["Not urgent", "Normal", "Urgent"]
      },
      "refund": {
        "type": "noul",
        "instructions": "Is the customer asking for a refund?"
      }
    }
  }'

Example response:

{
  "model": "laya",
  "answers": {
    "department": {
      "type": "choice",
      "choice": "billing",
      "confidence": 0.78,
      "probabilities": { "billing": 0.852, "account": 0.087, "technical": 0.061 }
    },
    "urgency": {
      "type": "score",
      "score": 1.2,
      "confidence": 0.34,
      "legend": { "0": "Not urgent", "1": "Normal", "2": "Urgent" },
      "probabilities": { "0": 0.12, "1": 0.56, "2": 0.32 }
    },
    "refund": { "type": "noul", "noul": 0.91 }
  },
  "routing": { "model": "laya:en" },
  "usage": { "input_tokens": 71, "output_tokens": 0 },
  "total_duration": 41250000,
  "load_duration": 0
}

/api/decide returns the TypeSafe response shape plus routing (the checkpoint the laya router picked) and timings in nanoseconds. See the API reference.

Question types

Type	`criteria`	Answer
`choice`	Object of option → description, up to 255 options	`choice`, `confidence`, `probabilities`
`score`	Ordered list of 2–10 levels	`score` (expected value), `confidence`, `legend`, `probabilities`
`noul`	Optional `{"true": "…", "false": "…"}`	`noul`: probability that the answer is true

confidence is the normalized top probability, (K · p_max − 1) / (K − 1) for K options: 0 when every option is equally likely, 1 when one option has all the probability.

Performance

Figures published on the Laya model card, measured on an NVIDIA Tesla T4:

	Latency, 1 question	Latency, 10 questions (batched)	Calibration error (ECE)
Laya (English)	39.5 ms	158.6 ms	0.081 after temperature fitting
Laya multilingual	32.8 ms	72.3 ms	—
TypeSafe Jev	236–276 ms p50 (third-party)	—	0.246

On batched calls laya:multilingual is up to ~2.2× faster than laya:en; for a single question the two are close. laya:typed-decisions reaches 0.766 accuracy on typed-decisions, against 0.727 published for Jev 1.13. Jev latencies come from AbdelStark/jev-benchmarks and nibzard/decision-model-benchmark; setups differ.

ONNX export

Ollaya runs Laya as ONNX. Across 2,383 questions per checkpoint (en, multilingual and typed-decisions), the ONNX export chose the same answer as the PyTorch fp32 reference 100% of the time, with a maximum probability difference of 1.1 × 10⁻⁴.

Limitations

Zero-shot typed decisions. The base checkpoints (en, multilingual) are near chance zero-shot on typed-decisions (0.362). Use laya:typed-decisions for those workflows.
Many options. Choice questions with more than ~20 options are weaker: 0.425 on Banking77, against 0.870 for Jev.
Calibration. The raw checkpoints are over-confident unless their temperatures are refit. For thresholds you rely on, refit calibration on your own labelled data and bake it in with CALIBRATION in a Modelfile.

License

Apache-2.0. Laya is developed by Convai Innovations — see the model card on Hugging Face. Ollaya packages the published weights for local serving.