by Tencent·Released 4 November 2026
This model accepts text and pictures as input and produces text as output. It can process up to 1000 pages of material at one time and generate responses up to 65 pages long. It is designed for complex, real-world productivity tasks that require planning and the use of specific tools. However, it cannot access the internet to find new information.
You pay for what each request uses. No subscription, no minimum.
50 % offTencent has cut its price. Every figure here is the new one.
One message, the ordinary use
Every rate
| Meter | ↑ you send | ↓ comes back |
|---|---|---|
| Text tokens | 138,79 DA | 416,21 DA |
| Cached tokens read | 6,99 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 | Cache read | Latency | Throughput | Uptime |
|---|---|---|---|---|---|---|
| Tencent50 % offAsia | … | … | … |
You pay the price of the provider that served your request. Prices read 15:06 · 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="tencent/hy4",
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_tokenstemperaturestopReasoning
Let it think before answering — and see the thinking.
reasoningreasoning_effortinclude_reasoningStructured output
Get JSON you can parse instead of prose.
response_formatstructured_outputsTools
Let the model call your functions.
toolstool_choiceOther
Also accepted by this model.
max_completion_tokensAbout 0,35 DA per message. Was 0,70 DA. ↑ 138,79 DA · ↓ 416,21 DA per 1M tokens.
No. You pay per use from your balance, with no subscription. Try it in Chat at the same price.
This model accepts text and pictures as input and produces text as output. It can process up to 1000 pages of material at one time and generate responses up to 65 pages long. It is designed for complex, real-world productivity tasks that require planning and the use of specific tools. However, it cannot access the internet to find new information.
Deposit dinars with your Edahabia or CIB card. Every model, Hy4 included, is paid from that balance. No foreign card, no conversion.
Tencent names no languages for Hy4. Arabic, French and Darja: not stated.
It reads text and images. Up to 1M tokens in one request, about 1 000 pages. It writes up to 49 000 words at once.
Yes, its weights are public. Open weights on Hugging Face
Tencent. Released 4 November 2026.
Hy Image 3.5 Preview takes a written description plus up to 20 reference pictures and returns an image up to 4K, or an edited version of a picture you supply. It can keep one subject looking the same across a series of edits, follows your description closely, and draws Chinese and English lettering inside the picture. You can edit over several rounds.
Hy4 preview reads and writes text, and keeps about 1,000 pages in mind at once. Tencent built it for coding assistants, long jobs that use the tools and apps you connect to it, and office-style work that needs planning and a steady thread over many steps. Pick it when the material is very long.
Hy3 reads and writes text, and keeps about 260 pages in mind at once. You choose how hard it thinks: not at all by default, or at a low or high level for tricky maths, coding and problems with many steps. Tencent points it at coding, game development, financial analysis, document work and web page design, and says it flags missing evidence rather than inventing an answer. Hy4 preview reads about 1,000 pages at once.
Hy3 preview reads and writes text, and keeps about 260 pages in mind at once. You can switch its thinking off, or set it to low or high, trading speed for depth depending on the job. Tencent builds it for long jobs with many steps and for writing code. As a preview it is for trying this line before settling on the finished Hy3, which has the same reading size.
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
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
A prompt it saw a moment ago costs less the second time.
Open weights
The model's weights are public.
Plan a complex project using provided tools
Thinks first · Uses your tools
Structure data into a specific format
Fills your JSON
Analyze a large document set and summarize findings
Cheaper on repeats · Open weights