opendunes

Ember-1

by FireworksReleased 24 September 2026

Ember-1 thinks before it answers, for coding, knowledge work and tasks it carries out on its own. It takes text and images, answers in text, and keeps about 1,000 pages in mind at once. Fireworks retrained Kimi K3 to think more briefly, cutting the length of its thinking by roughly 40 percent at comparable quality.

TakesTextImages
GivesText

Where your request runs

One company runs this model. OpenDunes sends every request to it. DA per 1M tokens: ↑ what you send · ↓ what comes back.

Fireworks…
↑ Sent, per 1M
998,51 DA
↓ Back, per 1M
4 992,53 DA
Cache read
99,85 DA
Latency
…
Throughput
…
Provider↑ Sent↓ BackCache readLatencyThroughputUptime
Fireworks998,51 DA4 992,53 DA99,85 DA………

You pay the price of the provider that served your request. Prices read 02:48 · speed over the last 30 minutes · uptime over 3 days.

What it costs

You pay for what each request uses. No subscription, no minimum.

≈ 3,49 DA per message3 000 DA buys about 850 messages

One message, the ordinary use

1 000 sent1 000 tokens sent× 998,51 per 1M× 998,51 DA per 1M1,00 DA
500 back500 tokens back× 4 992,53 per 1M× 4 992,53 DA per 1M2,50 DA
Total≈ 3,49 DA

Every rate

Meter↑ you send↓ comes back
Text tokens998,51 DA4 992,53 DA
Cached tokens read99,85 DA

per 1M tokens

Everyday jobs, priced

Answer a quick question≈ 1,40 DA
Read an invoice photo and pull out the total≈ 2,00 DA
Reply to a customer email≈ 2,60 DA
Explain a stack trace and suggest a fix≈ 4,49 DA
Review a 300-line pull request≈ 8,49 DA
When you pay
After the answer, from your balance. Nothing is held.
Why it varies
A longer question or a longer answer costs more.
After each one
The exact charge shows on the request.
Add creditDeposits from 100 DA, withEdahabiaCIB

See it in action

Real requests, real replies. Try any of them yourself — one click, pre-filled.

Understands images
You send

What's the total on this invoice, and when is it due?

invoice-scan.jpg · photoTry in Chat Lab
Structured output
You send

Turn this customer email into JSON with intent, urgency and order_id.

customer-email.txtTry in Chat Lab

Try your own request — same model, pre-selected.

Open Chat Lab

Call it in code

Swap the base URL and the key. Your OpenAI code runs unchanged.

POST /v1/chat/completions
Base URL
https://opendunes.com/v1
Model ID
fireworks/ember-1
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="fireworks/ember-1",
    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.

Parameters you can send

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_kstopfrequency_penaltypresence_penaltyrepetition_penaltylogit_biaslogprobstop_logprobs

Reasoning

Let it think before answering — and see the thinking.

reasoningreasoning_effortinclude_reasoning

Structured output

Get JSON you can parse instead of prose.

response_formatstructured_outputs

Tools

Let the model call your functions.

toolstool_choice

Limits

The hard numbers — what it can take in, what it can give back, and where it stops.

Context window
1 048 576 tokens
Max output
943 718 tokens
Reads at once
≈ 1 000 pages
Longest reply
≈ 708 000 words
Tokenizer
Other
Released
24 Sep 2026
Full API reference

Questions about Ember-1

How much does Ember-1 cost in dinars?

About 3,49 DA per message. ↑ 998,51 DA · ↓ 4 992,53 DA per 1M tokens.

Is Ember-1 free? Can I try it?

No. You pay per use from your balance in dinars, with no subscription. The chat has free models for free messages every day.

What is Ember-1 best for?

Ember-1 thinks before it answers, for coding, knowledge work and tasks it carries out on its own. It takes text and images, answers in text, and keeps about 1,000 pages in mind at once. Fireworks retrained Kimi K3 to think more briefly, cutting the length of its thinking by roughly 40 percent at comparable quality.

How do I pay for Ember-1 without an international card?

Add credit in dinars with your Edahabia or CIB card. Every model, Ember-1 included, is paid from that balance. No foreign card, no conversion.

Does Ember-1 work in Arabic, French or Darja?

Fireworks names no languages for Ember-1. Arabic, French and Darja: not stated.

Can it read PDFs, images or a whole book?

It reads text and images. Up to 1M tokens in one request, about 1 000 pages. It writes up to 708 000 words at once.

Is Ember-1 open source?

No. Fireworks has not published its weights.

Who makes Ember-1, and when was it released?

Fireworks. Released 24 September 2026.

Similar models, priced in dinars

See all models
Gemini Omni 1.2 FlashGoogle · Nov 5, 2026

This model is designed to handle text, pictures, and video inputs, allowing it to read up to 130 pages of content. It specializes in generating and editing video clips, producing outputs that include both text and moving images. What sets this model apart is its native ability to process and generate video content directly. It is not suitable for tasks requiring external tool usage or the creation of structured data formats.

Searches the webThinks first130 pages at once
≈2,00 DAper message
View model
DeepSeek V4.1 Flash 1017DeepSeek · Oct 17, 2026

This is the 17 October 2026 snapshot of V4.1 Flash, the image-and-text model DeepSeek serves under its current Flash name. It keeps about a thousand pages in mind, replies up to about 393 pages, thinks first, uses your connected tools and fills in the exact form you give it. Pick it for agents and coding work that include screenshots, and when you want the newest Flash. The earlier DeepSeek V4.1 Flash is the snapshot before this one; V4 Pro 0813 is the larger model for hard reasoning.

Thinks firstUses your tools1 000 pages at once
≈0,095 DAper message
View model
Gemini Omni 1.1 FlashGoogle · Nov 4, 2026

This model accepts text, pictures, and videos as input, allowing it to read up to 130 pages of content. It specializes in creating and editing video content, converting descriptions or pictures into moving images. A key feature is its ability to search the web for information while processing complex visual data. However, it does not utilize external tools or generate structured data formats.

Searches the webThinks first130 pages at once
≈2,00 DAper message
View model
Hy4Tencent · Nov 4, 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.

Thinks firstUses your tools1 000 pages at once
≈0,35 DAper message
View model