GPT-OSS 120B - AI Model API and Pricing

Volume work that suits an open-weights model

This is the larger of OpenAI's two open-weights gpt-oss models, listed for high-volume copy on a tight budget. Think catalog descriptions for a whole product range or first drafts of a weekly newsletter. Many headline variants for an A/B test fit too. So does placeholder text for a site you are still designing. You will edit the output either way, so a cheap first draft is a good trade.

Running it in AI Article Generator

It powers AI Article Generator, which offers six presets. The presets are SEO article, Blog post, Product description, Ad copy, Video script and Newsletter. Product description and Ad copy are the most forgiving, because short structured outputs hide the gap to stronger models. SEO article and Blog post work for drafts you will rewrite.

Feed it specifics. For a product line: "Write a three-sentence description for a merino wool hiking sock, size range S to XL. Mention cushioned heel, odor resistance and machine washing. Plain tone, no exclamation marks." Then swap the product details and run again.

How the weaker reasoning shows up

Its listed limit is that it is weaker than closed models. In practice, longer pieces drift: arguments repeat, sections overlap, and the conclusion may not follow from the body. Keep each run short and well bounded. State a word count in the prompt so it does not pad.

When to step up to another text model

DeepSeek V4 costs 2 credits per 1k words and is listed for long articles that need a coherent argument. It is the first upgrade for essays and analysis. Gemini 3 Flash, at 5 credits, is fast and good for its rate on high-volume captions. Claude Haiku 4.5, at 8 credits, writes natural sentences for headlines and ad variants where tone matters.

The full range for text runs from 1 to 47 credits per 1k words. This model sits first from the cheapest of 10.

What a batch really costs

An 800-word blog post rounds up to 1k words and costs 1 credit, about $0.01 at the best pack. The longer article in the cost table is 2 credits, about $0.02. At these rates, a test batch of fifty variants is cheaper than one image on most image models.

Where to actually run it

GPT-OSS 120B API

curl -X POST https://ugesi.com/api/generate \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-oss-120b",
    "prompt": "a product shot on white marble"
  }'

<?php
$ch = curl_init('https://ugesi.com/api/generate');
curl_setopt_array($ch, [
    CURLOPT_POST => true,
    CURLOPT_RETURNTRANSFER => true,
    CURLOPT_HTTPHEADER => [
        'Authorization: Bearer YOUR_API_KEY',
        'Content-Type: application/json',
    ],
    CURLOPT_POSTFIELDS => json_encode([
        'model' => 'gpt-oss-120b',
        'prompt' => 'a product shot on white marble',
    ]),
]);
$response = json_decode(curl_exec($ch), true);
curl_close($ch);
print_r($response);

type GenerateRequest = {
  model: string;
  prompt: string;
};
type GenerateResponse = {
  job_id: number;
  job_token: string;
  credits_held: number;
  status: "queued" | "succeeded";
  output_text?: string;
};

const body: GenerateRequest = {
  model: "gpt-oss-120b",
  prompt: "a product shot on white marble",
};

const resp = await fetch("https://ugesi.com/api/generate", {
  method: "POST",
  headers: {
    "Authorization": `Bearer ${process.env.UGESI_API_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify(body),
});
if (!resp.ok) throw new Error(`HTTP ${resp.status}`);
const job = (await resp.json()) as GenerateResponse;
console.log(job.job_id, job.status);

import requests

resp = requests.post(
    "https://ugesi.com/api/generate",
    headers={"Authorization": "Bearer YOUR_API_KEY"},
    json={
        "model": "gpt-oss-120b",
        "prompt": "a product shot on white marble",
    },
    timeout=60,
)
resp.raise_for_status()
print(resp.json())

{
  "model": "gpt-oss-120b",
  "prompt": "a product shot on white marble"
}

The API comes with every credit pack, from Lite to Enterprise, and uses the same balance as the website. API documentation and access →

License and terms of use

GPT-OSS 120B is released under the Apache License 2.0, a permissive license that allows commercial use. The full rules are in our Terms of service.

People also look for