SDAIA: ALLaM 2 7B Instruct

ALLaM-2-7b instruction tuned model by SDAIA

allam-2-7b

AvailableSDAIAChat

Input / Output: $0.05 / $0.08
Context: 32K
Max output: 4K
Modalities: Chat

About ALLaM 2 7B Instruct

ALLaM-2-7B is an instruction-tuned bilingual language model from the National Center for Artificial Intelligence at Saudi Arabia's SDAIA, built for Arabic language technology. The 7-billion-parameter model was trained from scratch on English and then mixed Arabic and English text. It suits Arabic and English chat, drafting, and research prototypes where a small open model is enough.

Where it works well

It was designed for Arabic from the start rather than adapted from an English model, which shows in Arabic generation and understanding.
English is retained: the maker staged pretraining so Arabic training did not erase English ability.
At 7 billion parameters it answers quickly and costs little to serve compared with large general models.
It uses the standard chat format and has an Apache 2.0 license on the published weights.

When to choose another model

The maker's card covers a 'preview' release and cautions about incorrect or biased output, so run your own safety checks.
It supports only text in Arabic and English; other languages, images, or tools are not part of the card.
A small model will trail large frontier models on hard reasoning, long code tasks, and complex agent work.

Getting started

Create API key
Create a key in Console, then use it with every model on the platform.

Send your first request

Copy the example for your language and run it against the endpoint.

Text to textChat Completions API

POST/v1/chat/completions

curl https://api.tokenlab.sh/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-xxx" \
  -d '{
    "model": "allam-2-7b",
    "messages": [
      {"role": "user", "content": "Hello!"}
    ]
  }'

Pricing

The Verified price applies to Verified, which costs less on most models. The Official price is the model maker's published price and applies to the more reliable Official route. Auto bills the route that completes the request.

Pricing

per 1M tokens

Verified price: —
Official price: Input$0.05per 1M tokensOutput$0.08per 1M tokens

	Official priceper 1M tokens	Verified priceper 1M tokens	Discount
Input	$0.05	—	—
Output	$0.08	—	—

Usage & activity

Success rate is the share of requests that completed. Latency is how long a full response takes; P95 means 95% of requests finished within that time.

Usage & availability

Last 24 hours

Requests
Success rate
P95 latency
Total tokens

Model performance

Last updated Nov 1, 2026, 11:22 AM UTC

30-day success rate: 97.1%
30-day requests: 100+
Last active: yesterday

Data is based on aggregate user requests, excluding status checks.

Open in Console

Open ALLaM 2 7B Instruct in Console with a prompt ready to edit or send.

Help me try allam-2-7b with a short message at /v1/chat/completions. Show the reply, latency, and cost.

Use cases

Arabic customer replies
Draft support answers and FAQs in Modern Standard Arabic, then have a person approve them before sending.
Bilingual summarization
Summarize English reports into Arabic, or Arabic articles into English, for internal briefings.
Arabic NLP research
Use it as a baseline for studies in classification, rewriting, or dialogue in Arabic.
Low-cost prototypes
Build a first version of an Arabic assistant before deciding whether a larger model is needed.

Prompt examples

لخّص هذه الفقرة في ثلاث نقاط واضحة ثم اقترح عنوانًا قصيرًا لها.

Translate this customer complaint into Arabic and write a polite two-sentence reply.

اشرح الفرق بين الفائدة البسيطة والمركبة بمثال عددي بسيط.

Cost calculator

Monthly input tokens1M tokens

Monthly output tokens0.5M tokens

Estimated monthly cost$0.09

FAQ

Who develops the ALLaM language models?

ALLaM is developed by the National Center for Artificial Intelligence (NCAI) at the Saudi Data and AI Authority (SDAIA). The weights are published on Hugging Face under the humain-ai and ALLaM-AI organizations.

Which languages does it handle?

Arabic and English. The maker trained on 4 trillion English tokens first, then 1.2 trillion mixed Arabic and English tokens, so it keeps English skills while adding Arabic. Other languages are not documented.

Is it a good general-purpose model?

It is a capable small bilingual chat model, not a frontier model. For Arabic drafting and Q&A it is a sensible low-cost start; for demanding reasoning or long code tasks, move up to a larger model.

Does the card describe this exact ID?

The official card is titled ALLaM-7B-Instruct-preview, the release this family comes from. This page's statements come from that card and the catalog entry, not from a card for the 2-7b label itself.

Can I use it commercially?

The Hugging Face card lists Apache 2.0. The card also says developers must evaluate safety before deployment and that output does not represent official NCAI or SDAIA statements.

How much does ALLaM 2 7B Instruct cost?

On TokenLab, ALLaM 2 7B Instruct costs Input $0.05 / Output $0.08 per 1M tokens. The pricing table above shows the full breakdown.

What are the context window and output limit of ALLaM 2 7B Instruct?

ALLaM 2 7B Instruct accepts up to 32,768 tokens of context and returns up to 4,096 tokens in one response.

Which endpoint should ALLaM 2 7B Instruct use?

Use https://api.tokenlab.sh/v1/chat/completions for ALLaM 2 7B Instruct. The request example below shows the matching code shape.

Which operations does ALLaM 2 7B Instruct support?

ALLaM 2 7B Instruct supports Text to text. Select an operation above to see its endpoint and request example.