InferenceNetStructured Output Model

Schematron V2 Small

A specialized extraction model that turns messy HTML into clean, schema-conforming JSON. Schematron V2 Small delivers the family's highest extraction quality — approaching frontier-model accuracy on complex schemas and long pages at a small fraction of the cost.

import OpenAI from "openai";
import { z } from "zod";
import { zodResponseFormat } from "openai/helpers/zod";

const openai = new OpenAI({
  baseURL: "https://api.inference.net/v1",
  apiKey: "<YOUR_API_KEY>",
});

// Define the shape you want back. There is no prompt — the schema is
// the only instruction the model needs.
const Listing = z.object({
  title: z.string(),
  price: z.number(),
  city: z.string(),
});

// Messy HTML in, conforming JSON out.
const html = `<div class="listing">
  <h2>Brompton C Line</h2>
  <span class="p">£1,795</span>
  <p>Used · London</p>
</div>`;

const response = await openai.chat.completions.parse({
  model: "inference-net/schematron-v2-small",
  messages: [{ role: "user", content: html }],
  response_format: zodResponseFormat(Listing, "listing"),
  temperature: 0,
});

console.log(response.choices[0].message.parsed);
// { title: "Brompton C Line", price: 1795, city: "London" }

import OpenAI from "openai";
import { z } from "zod";
import { zodResponseFormat } from "openai/helpers/zod";

const openai = new OpenAI({
  baseURL: "https://api.inference.net/v1",
  apiKey: "<YOUR_API_KEY>",
});

// Define the shape you want back. There is no prompt — the schema is
// the only instruction the model needs.
const Listing = z.object({
  title: z.string(),
  price: z.number(),
  city: z.string(),
});

// Messy HTML in, conforming JSON out.
const html = `<div class="listing">
  <h2>Brompton C Line</h2>
  <span class="p">£1,795</span>
  <p>Used · London</p>
</div>`;

const response = await openai.chat.completions.parse({
  model: "inference-net/schematron-v2-small",
  messages: [{ role: "user", content: html }],
  response_format: zodResponseFormat(Listing, "listing"),
  temperature: 0,
});

console.log(response.choices[0].message.parsed);
// { title: "Brompton C Line", price: 1795, city: "London" }

How it worksSchema-constrained decoding — output is valid JSON by construction, not by retry.

Input · raw HTML

<div class="listing">
  <h2>Brompton C Line</h2>
  <span class="p">£1,795</span>
  <p>Used · London</p>
</div>

Input · JSON Schema

{ "title":  "string",
  "price":  "number",
  "city":   "string" }

schematron-v2-small

Output · conforming JSONAlways valid

{
  "title": "Brompton C Line",
  "price": 1795,
  "city":  "London"
}

Related models

Model

InputCache readOutput

Kimi K3

1024K input contextVisionToolsReasoning$3.00 / $15.00

1000K input contextTools$2.50 / $5.00

$2.50

$5.00

Schematron V2 TurboInference.net

125K input context$0.03 / $0.15

$0.03

$0.15

Start building with Schematron V2 Small today

Serverless, OpenAI-compatible, and billed per token. No commitments.