by xAI·Released 12 August 2026
#2 for mobile apps · Coding, knowledge work and science and maths reasoning over long inputs, with tool calls and JSON output
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 | 532,54 DA | 1 597,61 DA |
| Cached tokens read | 133,13 DA |
per 1M tokens
Web search · 1,66 DA per call, under every choice
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 |
|---|---|---|---|---|---|---|
| xAI | 532,54 DA | 1 597,61 DA | 133,13 DA | … | … | … |
Also charged: prompts over 200 000 tokens ↑ 1 065,07 ↓ 3 195,22 · a web search 1,66 DA each.
You pay the price of the provider that served your request. Prices read 15:18 · speed over the last 30 minutes · uptime over 3 days.
Swap 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="x-ai/grok-4.6:batch",
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_pseedlogprobstop_logprobsReasoning
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_choiceAbout 1,33 DA per message. ↑ 532,54 DA · ↓ 1 597,61 DA per 1M tokens.
No. You pay per use from your balance, with no subscription. Run it once from Call it in code, at the same price.
#2 for mobile apps · Coding, knowledge work and science and maths reasoning over long inputs, with tool calls and JSON output.
Deposit dinars with your Edahabia or CIB card. Every model, Grok 4.6 (batch) included, is paid from that balance. No foreign card, no conversion.
xAI names no languages for Grok 4.6 (batch). Arabic, French and Darja: not stated.
It reads text, images and PDFs. Up to 500K tokens in one request, about 500 pages. It writes up to 338 000 words at once.
No. xAI has not published its weights.
xAI. Released 12 August 2026.
Grok Voice TTS 1.0 reads text aloud in 20 or more languages and detects the language by itself. Its built-in voices are named Eve, Ara, Rex, Sal and Leo, and short tags in the text set speed, pitch, pauses, emphasis and style. You get a sound file such as MP3 or WAV, with at most 15,000 characters at a time. It only speaks; Grok STT 1.0 does the reverse and writes speech down.
The 2.0 release of Grok Imagine Image, turning prompts into pictures and editing from up to three references at 1K or 2K, in low or medium quality mode; the Quality model is its photorealistic sibling
Grok STT 1.0 listens to a recording and writes down what was said, with a time for every word. It can mark who is speaking and handles recordings with several audio channels. Pick it when you need a written record of a recording with times and speakers. It only turns speech into text; Grok Voice TTS 1.0 does the reverse and reads text aloud.
Grok Imagine Video 1.5 makes a video from a written description, with an optional starting picture to set the scene. You can direct how the subject and camera move, the pacing and the mood, and it can add matching sound effects, background sound and spoken lines. Pick it for the best quality of the three. Grok Imagine Video 1.5 Lite is quicker, and Grok Imagine Video takes up to seven reference pictures.
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
Can look things up online while it answers.
Cheaper on repeats
A prompt it saw a moment ago costs less the second time.