What's the total on this invoice, and when is it due?
invoice-scan.jpg · photoThe total is 42 300,00 DA, due 5 September 2026. Worth a double-check: the handwritten day could read as 8 instead of 3.
Try in Chat Labby Perceptron·Released 12 May 2026
Perceptron Mk1 reads a written question together with pictures or video and writes back, keeping about 32 pages in mind at once. It is built for video and for understanding the physical world: answering questions about a clip, summarising it and spotting events, plus reading text in messy documents, counting and finding objects by name, and hand positions in pictures. Ask for points, boxes or outlines on pictures, or start and end times in video. Perceptron Mk1.5 adds sound input and connected tools.
You pay for what each request uses. No subscription, no minimum.
One message, the ordinary use
Every rate
| Meter | ↑ you send | ↓ comes back |
|---|---|---|
| Text tokens | 49,93 DA | 499,25 DA |
per 1M tokens
One company runs this model. OpenDunes sends every request to it. DA per 1M tokens: ↑ what you send · ↓ what comes back.
| Provider | ↑ Sent | ↓ Back | Latency | Throughput | Uptime |
|---|---|---|---|---|---|
| Perceptron | 49,93 DA | 499,25 DA | … | … | … |
You pay the price of the provider that served your request. Prices read 15:48 · speed over the last 30 minutes · uptime over 3 days.
Real requests, real replies. Try any of them yourself — one click, pre-filled.
What's the total on this invoice, and when is it due?
invoice-scan.jpg · photoThe total is 42 300,00 DA, due 5 September 2026. Worth a double-check: the handwritten day could read as 8 instead of 3.
Try in Chat LabTurn this customer email into JSON with intent, urgency and order_id.
customer-email.txtTry your own request — same model, pre-selected.
Open Chat LabSwap the base URL and the key. Your OpenAI code runs unchanged.
import os
from openai import OpenAI
client = OpenAI(
base_url="https://opendunes.com/v1",
api_key=os.environ["OPENDUNES_API_KEY"],
)
resp = client.chat.completions.create(
model="perceptron/perceptron-mk1",
messages=[{"role": "user", "content": "Two lines to welcome a shop in Oran."}],
)
print(resp.choices[0].message.content)Runs this request once on your balance.
Standard OpenAI fields, grouped by what they control. Anything not listed is accepted and ignored.
Shaping the reply
How long, how varied, where to stop.
max_tokenstemperaturetop_ptop_kfrequency_penaltypresence_penaltyReasoning
Let it think before answering — and see the thinking.
reasoninginclude_reasoningStructured output
Get JSON you can parse instead of prose.
structured_outputsAbout 0,30 DA per message. ↑ 49,93 DA · ↓ 499,25 DA per 1M tokens.
No. You pay per use from your balance, with no subscription. Try it in Chat at the same price.
Perceptron Mk1 reads a written question together with pictures or video and writes back, keeping about 32 pages in mind at once. It is built for video and for understanding the physical world: answering questions about a clip, summarising it and spotting events, plus reading text in messy documents, counting and finding objects by name, and hand positions in pictures. Ask for points, boxes or outlines on pictures, or start and end times in video. Perceptron Mk1.5 adds sound input and connected tools.
Deposit dinars with your Edahabia or CIB card. Every model, Perceptron Mk1 included, is paid from that balance. No foreign card, no conversion.
Perceptron names no languages for Perceptron Mk1. Arabic, French and Darja: not stated.
It reads text, images and video. Up to 33K tokens in one request, about 32 pages. It writes up to 6 000 words at once.
No. Perceptron has not published its weights.
Perceptron. Released 12 May 2026.
Perceptron Mk1.5 is made for physical agents that must understand what they see and hear. You send text, pictures, video or audio, and it replies in text, with optional points, boxes, outlines, tracks or clips marking what it found if you ask. It can use the tools you connect, fills in the exact form you give it, lets you set how hard it thinks from minimal to high, and listens to a video's soundtrack only when you switch that on. Perceptron Mk1 has no audio input.
From what its maker and two benchmark sites publish, and where it stands on OpenDunes. Read 7 Nov 2026.
Thinks first
Works through the problem before it answers.
Uses your tools No.
Can call functions your code provides.
Fills your JSON
Answers in the JSON shape you give it.
Searches the web No.
Can look things up online while it answers.
Cheaper on repeats No.
A prompt it saw a moment ago costs less the second time.
Open weights No.
Read on docs.perceptron.inc.
Summarise a security camera clip
Video
Detect when a delivery arrives in a video
Video
Read a crumpled receipt photo into JSON
Images · Fills your JSON