NeoSyntropy
NEOSYNTROPY / THE AI RUNTIME ENGINE NS / 001

Control what AI does
next, at runtime.

NeoSyntropy is a runtime engine that controls language models through finite state machines. Every action follows a path you define—your business logic is enforced by the engine, not left to a prompt.

Make your first API call ↗Explore the nodes ↓
01 / VALIDATED JSONAI outputs match the shape your application expects.02 / RUNTIME LOGICEach action moves through a defined state and transition.03 / APPROVED PATHSInvalid output stops before it creates a side effect.
API EXPLORER / 01ILLUSTRATIVE EXAMPLE
SCHEMA API

Unstructured input. A defined output.

INPUT

Refund $40 for order ORD-204. The item arrived damaged.

{ }Schema API
OUTPUTCONFIDENCE 0.96
{
  "order_id": "ORD-204",
  "amount": 40,
  "reason": "damaged",
  "confidence": 0.96,
  "out_of_distribution": false
}
OOD CHECK · PASS

Illustrative decoded output. Your application can reject low-confidence or out-of-distribution results before they continue.

POST /api/v1/schemaYOUR TASK. DEFINED BOUNDARIES.

01 / ACCESS LAYER

One intelligence layer.
Three ways in.

Start with a single API call, compose a workflow, or connect the agent you already run.

01 / CALL INTELLIGENCE

NeoSyntropy API

Schema output. Reasoning steps. Routing decisions. Call the capability you need directly over HTTP.

Explore the API
02 / COMPOSE INTELLIGENCE

NeoSyntropy Framework

Build with typed AI nodes. Connect reasoning, tools, and routing inside a graph you define.

Build a workflow
03 / CONNECT INTELLIGENCE

NeoSyntropy Proxy

Send model calls from existing agents through one gateway. Keep your runtime and centralize provider access.

Connect your agent

01A / INPUT TOKEN PRICING

Pay for input tokens.
Not model size.

You pay for the input tokens sent to NeoSyntropy. A model’s parameter count describes its capacity—it does not change your usage bill.

See full pricing ↗
STANDARD INPUT$0.10per 1M input tokens
TUNED INPUT$0.05per 1M input tokens
STARTER CREDIT$10minimum deposit / year

01B / API INPUT PRICING

Intelligence at
production scale.

NeoSyntropy API input pricing starts at $0.10 per million tokens. The comparison below is a simple input-cost illustration; provider prices and model availability can change.

Illustrative input cost per 1 million tokens
Model pathInput costRelative to NeoSyntropy
NeoSyntropy API$0.101×
GPT-5.6 Luna$0.202×
Claude Haiku 4.5$1.0010×
GPT-5.6 Terra$2.0020×
Claude Sonnet 5$10.00100×
100×lower input cost than the $10.00 comparison path

02 / INTELLIGENCE PRIMITIVES

A defined job.
An intelligent node.

{ }01 / SCHEMA NODE

Give language
a structure.

Extract fields from messages and documents against a schema your application defines.

Explore schema output ↗
∴02 / REASONING STEP NODE

Give reasoning
a responsibility.

Supply context and permitted tools. Inspect the result of each step before the workflow continues.

Explore reasoning steps ↗
â‘‚03 / ROUTER

Give decisions
a destination.

Let AI choose among defined candidates. Add deterministic rules where the policy must be exact.

Explore routing ↗

03 / THE EXECUTION LAYER

You define the map.
The runtime
checks the path.

Models propose. Your graph defines allowed transitions. Schemas, guards, and execution traces make each step inspectable.

Inside the framework ↗
WORKFLOW / REFUND REVIEW EXAMPLE
01
SchemaNodeExtract the request
02
SemanticRouterSelect an approved path
03
ReasoningNodeGather evidence with tools
04
Transition guardCheck policy before proceeding
DEFINED PATH → INSPECTABLE EXECUTION

BUILD SOMETHING THAT WORKS.

Your next AI system
starts with one node.

Explore the API ↗Framework quickstart

Connecting an existing agent? Meet NeoSyntropy Proxy →