Summarize

Overview

Summarize is called through the OpenAI-compatible POST /v1/chat/completions endpoint. Set model to summarize (latest) or summarize-1 (pinned version) and send the text as the last user message. The model condenses it in its own words without adding new information or commentary. Unlike extractive summarization, the output is a fluent rewrite - not a selection of lifted sentences. Because the endpoint is OpenAI-compatible, you can point an unmodified OpenAI or OpenRouter SDK at ScaleDown by changing only the base URL and API key.

POST https://api.scaledown.xyz/v1/chat/completions

Request

string

required

"summarize" (always the latest version) or "summarize-1" (pinned to version 1). Both currently behave identically.

array

required

Standard OpenAI chat messages. The last user message is the text to summarize. Any system messages are joined and used as the summary instructions.

number

default:1000

Standard OpenAI field. Maximum number of tokens in the generated summary. Defaults to 1000, which is also the maximum: larger values are capped at 1000, and the model is instructed to finish within the limit. If a summary still hits the limit it is cut off and the parsed result sets finish_reason to "length" with a notice. For a higher output limit, email neal@scaledown.ai.

string

Non-standard extension, passed with the SDK’s extra_body. A base64-encoded file to summarize. Supported formats: JPEG, PNG, TIFF, single-page PDF, multi-page PDF. The text is read from the file and used as input.

string

Non-standard extension, passed with extra_body. MIME type of the document (e.g. "image/jpeg", "application/pdf"). Required when document is provided.

string

Optional top-level override for the input text. When present it takes precedence over the last user message - this lets a batch item body be reused verbatim. Likewise a top-level instructions string takes precedence over system messages.

Temperature, top-p, and other sampling parameters are fixed and not configurable.

Response

A standard chat.completion. The summary result is a JSON string in choices[0].message.content - parse it to read the fields below.

string

Echoes the model you sent (summarize or summarize-1).

object

prompt_tokens is the input tokens billed for the request, completion_tokens the model’s output tokens, and total_tokens their sum.

string

A JSON string with these fields:

Show Parsed content

string

The generated summary text.

number

Character count of the input text.

number

Character count of the generated summary.

number

End-to-end request latency in milliseconds.

number

Input tokens billed for the request. Output length is capped at 1000 tokens and does not add to the input-token charge.

string

"stop" when the summary finished on its own, or "length" when it was cut off at the output limit.

string | null

Present only when finish_reason is "length". Explains that the output limit was reached and that a higher limit is available by contacting neal@scaledown.ai.

string | null

The raw text extracted from the document. null if no document was provided.

Error responses

Status	Meaning
`400 Bad Request`	No input text (no user message, `text`, or `document`).
`402 Payment Required`	Insufficient credits.
`422 Unprocessable Entity`	Malformed request body or the document could not be read.
`500 Internal Server Error`	Inference service unavailable.
`504 Gateway Timeout`	Summarization request timed out.

Authentication

Include your API key in every request using the x-api-key header. Authorization: Bearer <your-api-key> is also accepted, so an OpenAI SDK only needs its api_key set.

-H "x-api-key: <your-api-key>"

Examples

Basic summary

curl -X POST https://api.scaledown.xyz/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "x-api-key: <your-api-key>" \
  -d '{
    "model": "summarize",
    "messages": [
      {"role": "user", "content": "Your long text here..."}
    ]
  }'

import json
from openai import OpenAI

client = OpenAI(
    base_url="https://api.scaledown.xyz/v1",
    api_key="<your-api-key>",
)

r = client.chat.completions.create(
    model="summarize",
    messages=[{"role": "user", "content": "Your long text here..."}],
)
print(json.loads(r.choices[0].message.content)["summary"])

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.scaledown.xyz/v1",
  apiKey: "<your-api-key>",
});

const r = await client.chat.completions.create({
  model: "summarize",
  messages: [{ role: "user", content: "Your long text here..." }],
});
const result = JSON.parse(r.choices[0].message.content!);
console.log(result.summary);

Response:

{
  "id": "chatcmpl-9f2c1e0a7b3d4c5e8a6b1d2f3e4a5b6c",
  "object": "chat.completion",
  "created": 1760000000,
  "model": "summarize",
  "choices": [{
    "index": 0,
    "message": {
      "role": "assistant",
      "content": "{\"summary\":\"The company reported strong Q3 results, approved a share buyback program, and announced plans for Southeast Asian expansion.\",\"input_chars\":8340,\"output_chars\":142,\"latency_ms\":3241,\"input_tokens\":2105,\"finish_reason\":\"stop\",\"notice\":null}"
    },
    "finish_reason": "stop"
  }],
  "usage": { "prompt_tokens": 2105, "completion_tokens": 38, "total_tokens": 2143 }
}

With instructions

Put format, language, focus area, or length guidance in a system message. It extends, and does not replace, the default behaviour (no new information, no commentary).

curl -X POST https://api.scaledown.xyz/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "x-api-key: <your-api-key>" \
  -d '{
    "model": "summarize-1",
    "max_tokens": 150,
    "messages": [
      {"role": "system", "content": "Use bullet points. Focus on dates and key decisions only."},
      {"role": "user", "content": "Your long text here..."}
    ]
  }'

r = client.chat.completions.create(
    model="summarize-1",
    max_tokens=150,
    messages=[
        {"role": "system", "content": "Use bullet points. Focus on dates and key decisions only."},
        {"role": "user", "content": "Your long text here..."},
    ],
)
print(json.loads(r.choices[0].message.content)["summary"])

const r = await client.chat.completions.create({
  model: "summarize-1",
  max_tokens: 150,
  messages: [
    { role: "system", content: "Use bullet points. Focus on dates and key decisions only." },
    { role: "user", content: "Your long text here..." },
  ],
});
console.log(JSON.parse(r.choices[0].message.content!).summary);

The parsed content is:

{
  "summary": "- The company reported Q3 revenue of $4.2B, up 12% year-over-year.\n- The board approved a $500M share buyback program on October 14.\n- CEO announced plans to expand into Southeast Asia by mid-2027.",
  "input_chars": 8340,
  "output_chars": 198,
  "latency_ms": 3241,
  "input_tokens": 2105,
  "finish_reason": "stop",
  "notice": null
}

Summarizing a document

Pass a base64-encoded image or PDF through extra_body (extra_body fields are merged into the top level of the request). The text is extracted automatically and summarized, and the raw extracted text is returned as ocr_text. A message is still required; use a short placeholder user message.

DOCUMENT=$(base64 -b 0 -i report.pdf)

curl -X POST https://api.scaledown.xyz/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "x-api-key: <your-api-key>" \
  -d '{
    "model": "summarize",
    "messages": [
      {"role": "system", "content": "Use bullet points. Focus on key figures and decisions only."},
      {"role": "user", "content": "Summarize the attached document."}
    ],
    "document": "'"$DOCUMENT"'",
    "document_mime_type": "application/pdf"
  }'

import base64

with open("report.pdf", "rb") as f:
    document = base64.b64encode(f.read()).decode()

r = client.chat.completions.create(
    model="summarize",
    messages=[
        {"role": "system", "content": "Use bullet points. Focus on key figures and decisions only."},
        {"role": "user", "content": "Summarize the attached document."},
    ],
    extra_body={"document": document, "document_mime_type": "application/pdf"},
)
print(json.loads(r.choices[0].message.content)["summary"])

import fs from "fs";

const document = fs.readFileSync("report.pdf").toString("base64");

const r = await client.chat.completions.create({
  model: "summarize",
  messages: [
    { role: "system", content: "Use bullet points. Focus on key figures and decisions only." },
    { role: "user", content: "Summarize the attached document." },
  ],
  // @ts-expect-error ScaleDown extension fields
  document,
  document_mime_type: "application/pdf",
});
console.log(JSON.parse(r.choices[0].message.content!).summary);

Notes

The base instructions (no new information, no commentary) are always applied. System messages add on top of them - they do not replace them.
GET /v1/models lists summarize and summarize-1. The bare name always points at the latest version; use summarize-1 to pin.
Streaming is not supported; the full result is returned in one response.

Extract

Classify

Compress

Summarize

Overview

Request

Response

Error responses

Authentication

Examples

Basic summary

With instructions

Summarizing a document

Notes

​Overview

​Request

​Response

​Error responses

​Authentication

​Examples

​Basic summary

​With instructions

​Summarizing a document

​Notes

Overview

Request

Response

Error responses

Authentication

Examples

Basic summary

With instructions

Summarizing a document

Notes