TypeSafe ASCII Decision Orb
Non-Autoregressive Decision Inference

We took the opposite research direction.

Generative models predict text sequentially token by token, accumulating latency, cost, and hallucination. System One uses a bidirectional encoder with direct decision heads to produce calibrated probabilities and structured choices in a single forward pass.

~33 ms
Inference Latency

Single-pass forward evaluation on batch workloads without token decoding delay.

322M
Backbone Weights

Bidirectional mmBERT-base encoder coupled with specialized decision heads.

100+
Languages

Native multilingual representation across Latin and non-Latin character sets.

0.0%
Hallucination

Direct classification heads eliminate generative drifting and schema corruption.

Foundation & Specification
b.64 VHlwZVNBRkUgU1lTVEVNIE9ORQ==

What are System One Models and Jev?

The System One paradigm was introduced by TypeSafe AI (founded by Diogo Almeida, former OpenAI research lead behind ChatGPT instruction-following). Inspired by Daniel Kahneman's distinction in Thinking, Fast and Slow, System One models deliver fast, intuitive, and mathematically calibrated decisions rather than slow, open-ended generative string completions.

Concept

System One vs Generative LLMs

Traditional LLMs predict one token at a time in recurrent loops. They are flexible for chat, but introduce high latency, type drift, and hallucination risks into automated code. System One models trade away string generation to execute parallel decisions directly over program state with zero type errors and zero hallucinations.

Origin: Daniel Kahneman, Thinking, Fast and Slow
TypeSafe Frontier Model

Jev & The Jevons Paradox

Jev is TypeSafe AI's flagship frontier model. Named after economist William Stanley Jevons, it embodies the thesis that dropping the cost and latency of decision intelligence by two orders of magnitude unlocks massive demand for automated software workflows, real-time routing, and deterministic guardrails.

Drop-in SDK & API Compatibility

Compatible with TypeSafe Jev client libraries and OpenAPI schemas

POST /v1/systemone

This server implements the exact TypeSafe Jev API contract. You can point any TypeSafe-compatible SDK or standard HTTP client to this self-hosted endpoint simply by changing the base URL:

noul Question
Binary truth probability [0.0 - 1.0] with calibrated confidence.
choice Question
Categorical routing with exact probability distribution across options.
score Question
Ordinal rubric evaluation returning float score, legend map, and confidence.
# Python example (Drop-in with TypeSafe API contract)
import httpx

response = httpx.post(
    "http://localhost:8000/v1/systemone",
    json={
        "state": "Payment failed for customer #4421.",
        "model": "laya",
        "questions": {
            "is_urgent": {"type": "noul", "instructions": "Is this time sensitive?"},
            "dept": {"type": "choice", "instructions": "Target team", "criteria": {"billing": "Billing", "tech": "Tech"}},
            "severity": {"type": "score", "instructions": "Incident severity", "criteria": ["Low", "High"]}
        }
    }
)
print(response.json()["answers"])
Live Tester

Interactive Playground

Execute live inference requests directly against the local POST /v1/systemone endpoint.

Presets:

Stored in session storage. Leave blank for public access (rate limited to 10 requests per minute per IP), or enter your configured API key for unlimited access.

Context string evaluated across all configured questions.

Map of question keys to question objects (types: noul, choice, score).

Evaluation Output
S1

Ready for evaluation.

Select a preset or enter input state and click Run System 1 Evaluation.

Model Card & Attribution

Laya Multilingual Decision Backbone

System One runs on Laya, a non-autoregressive decision model developed by Convai Innovations and Nandha Kishor. Laya replaces generative token prediction with direct probabilistic classification heads.

Project Credits, Author & Upstream Repositories
Web & Public API
Zulfi Fazhar

FastAPI service wrapper, TypeSafe Jev specification compatibility, and developer playground interface.

GitHub
Model Author
Nandha Kishor

Laya model architecture design, pretraining, and RLCD alignment implementation.

GitHub
Model Research
Convai Innovations

Research laboratory specializing in efficient decision-first architectures.

Hugging Face
Model Hub
convaiinnovations/laya

Official model checkpoints, tokenizer configurations, and weight releases.

Architecture Specifications
Backbone Architecture
mmBERT-base

Bidirectional encoder with 22 layers, hidden dimension 768, and 256k vocabulary.

Total Parameters
322 Million (322M)

Compact footprint suitable for local edge execution and high-density server nodes.

Context Window
1024 Tokens

Up to 1024 tokens context per question with 256 tokens dedicated to instructions.

Training Method
RLCD Alignment

Reinforcement Learning from Calibrated Decisions across 15,987 updates and 4 epochs.

Multilingual Scope
100+ Languages

Evaluated across 51 languages in the MASSIVE benchmark with superior accuracy on non-Latin scripts.

Inference Throughput
Up to 332 QPS

Batched evaluation throughput on standard hardware with ~32.8 ms single forward pass.

Open Source License
Apache 2.0

Permissive commercial and research license for weights and deployment code.

Technical Questions

Frequently Asked Questions

How does System 1 operate without autoregressive decoding?

Standard generative models predict text token by token in a recurrent loop. System One uses a bidirectional encoder backbone coupled with non-autoregressive decision heads. Both the state context and question instructions are processed in a single forward pass, producing deterministic classifications and calibrated probabilities without token decoding delays.

How are the output probabilities calibrated?

Probabilities are calibrated through temperature scaling and Reinforcement Learning from Calibrated Decisions (RLCD). When System One returns a probability of 0.85 for a binary decision or choice category, that classification is empirically accurate approximately 85% of the time across test distributions.

How do I integrate System One into an existing backend?

Send an HTTP POST request to the /v1/systemone endpoint with your state context and questions schema. The endpoint responds with structured JSON containing typed answers and token usage.

curl -X POST "http://localhost:8000/v1/systemone" \
  -H "Content-Type: application/json" \
  -d '{
    "state": "Customer charged twice for order #8841. Requesting full refund.",
    "questions": {
      "is_refund": {
        "type": "noul",
        "instructions": "Is the customer requesting a refund?"
      }
    }
  }'

What question types are supported?

System One supports three structured question types:

  • noul: Binary decision returning a calibrated float between 0.0 and 1.0.
  • choice: Categorical selection among named criteria options with probability distribution.
  • score: Ordinal rating across ranked criteria levels with score index and confidence.