Institute of Foundation Models and Inception42 Launch Jais 3: The World's Leading Arabic-Native AI for Speech, Writing and Agents

Cerebras Systems (NASDAQ: CBRS) collaborated with the Institute of Foundation Models and Inception42 on the newly launched Jais 3.

Rhea-AI Summary

Cerebras Systems (NASDAQ: CBRS) collaborated with the Institute of Foundation Models and Inception42 on the newly launched Jais 3. The Arabic AI fleet combines six open language models, ranging from 900 million to 375 billion parameters, with Jais Speech, an Arabic speech recognition model.

Built on IFM's K2 Horizon models, Jais 3 supports spoken and written Arabic across dialects and can use software tools to produce documents and complete multi-step tasks. Jais 3 and Jais Speech are available on Hugging Face, with deployment support for vLLM and SGLang. The models also power the Jais chat and mobile apps; a Jais platform for multi-step workflows is coming soon.

Key Figures

Language models: 6 models Model size range: 900 million to 375 billion parameters Arabic benchmarks: 24 benchmarks +3 more

Language models

6 models

Jais 3 open language model fleet

Model size range

900 million to 375 billion parameters

Six Jais 3 language models

Arabic benchmarks

24 benchmarks

Audited evaluation suite

Arabic evaluation average score

56.4 to 79.6

7-billion-parameter K2 Horizon base model vs. Jais 3

Context window

512,000 tokens

Jais 3 language models

Dialect coverage

19 Arab countries

Jais Speech transcription

Key Terms

context window, word and character error rates

2 terms

context window technical

"context window of 512,000 tokens"

The context window is the amount of text an AI system can read and consider at one time — like the size of a desk where it lays out documents to make sense of them. For investors, a larger context window means the AI can analyze longer filings, full earnings calls or combined news streams without losing earlier details, improving the accuracy of summaries, risk signals and investment decisions.

word and character error rates technical

"lowest average word and character error rates of the models compared"

Word error rate (WER) and character error rate (CER) are accuracy metrics for automated text outputs such as speech recognition, OCR, or machine translation. They measure the minimum number of edits—substitutions, deletions and insertions—needed to change the system output into a human reference, divided by the number of words (for WER) or characters (for CER) in that reference. WER reports errors at the word level and is sensitive to tokenization, spacing and compound words; CER measures errors at the character level and is more granular for short words or languages with different word boundaries. Both depend on the chosen reference transcription and on preprocessing rules (e.g., case, punctuation, normalization), which can materially affect the reported score.

AI-generated analysis. How Rhea-AI works. Not financial advice.

See more from StockTitan in Google Search and AI answers. Adds StockTitan as a preferred source · opens Google

Add on Google Not now

Built on IFM's K2 Horizon Fleet, Jais 3 Combines Leading Arabic AI Performance with Deep Cultural Understanding to Power Speech, Writing, and Agents

ABU DHABI, UAE, Oct. 15, 2026 /PRNewswire/ -- The Institute of Foundation Models (part of Mohamed bin Zayed University of Artificial Intelligence) and Inception42 today introduced Jais 3, a fleet of models built for Arabic. Jais 3 understands spoken and written Arabic across dialects, writes naturally in Arabic, and performs practical tasks for users, from transcribing conversations to producing presentations. The new fleet pairs six open language models, from 900 million to 375 billion parameters, with Jais Speech, a speech recognition model for Arabic and its dialects. All are built on top of K2 Horizon, IFM's latest fleet of fully open-source foundation models launched last month.

Jais 3 is trained to understand Modern Standard Arabic and the regional dialects people actually speak, from the Gulf to North Africa, along with the culture around the language, including its poetry, its cuisine and Islamic knowledge. For the more than 400 million people who speak Arabic, and for the hospitals, companies and government offices across the Middle East and North Africa that work in it, that understanding decides whether an AI tool can be trusted with everyday work.

"Arabic AI needs to understand the language in its everyday context, which varies by geography and across generations," said Haonan Li, senior research scientist at IFM. "For example, we need to account for the many nuances in spoken Arabic, including when speakers slip into English or French mid-sentence and back again. We built Jais 3 for Arabic as it's really used."

"Jais 3 makes Arabic AI more practical for business use. Stronger reasoning and instruction following, combined with tool use, provide a foundation for multi-step agentic workflows. With open models across a range of sizes, organisations can match these capabilities to their computing budgets and deployment needs," said Dr. Larry Murray, SVP Applied Science at Inception42.

IFM and Inception42 tested the new models on its own audited suite of 24 Arabic benchmarks. The largest Jais 3 model matches the best open models on Arabic understanding and leads them on agentic tasks, and Jais Speech leads every model compared on Arabic speech recognition. The advantage holds across sizes: the 3.7-billion-parameter model outperformed every competing open model of up to 5 billion parameters on Arabic and agentic tasks, and the 7-billion- and 32-billion-parameter models earned the highest average Arabic scores among open models tested at their sizes.

Jais Speech transcribes Modern Standard Arabic and dialects spoken across 19 Arab countries in noisy, real-world conditions. It follows speakers as they switch between Arabic and languages such as English or French, keeping those words in their original script. It also identifies the dialect being spoken and can add Arabic vowel marks to a transcript. In IFM's and Inception42's evaluation across six public test sets, Jais Speech recorded the lowest average word and character error rates of the models compared, including commercial speech recognition services. Paired with the language models, Jais Speech lets Jais 3 take spoken instructions in Arabic or English and act on them.

General-purpose models often drift between dialects and registers within a document. Jais 3 stays fluent and consistent throughout and follows detailed instructions closely. Given four instructions for an Arabic hospital appointment letter, including a word to avoid, it met all four; a general-purpose model used the banned word twice. In IFM's and Inception42's evaluations, the small and mid-size Jais 3 models lead the open models compared on following complex Arabic instructions.

Jais 3 also carries out work. Connected to software tools, it can plan multi-step tasks, search the web, and read, edit, and produce Arabic documents, spreadsheets, presentations, and websites. Asked for an Arabic presentation on the forecast for 2026–2027, El Nino, it researched the topic and delivered a four-slide deck built from published forecast data. The same models power the Jais chat app and mobile app, and a Jais agentic platform for everyday workflows is coming soon.

"Jais 3 brings deep Arabic understanding to speech, writing and agents," said Richard Morton, vice president and executive director of IFM. "It is built for the way Arabic is spoken, the way it is written, and the work people need done in it."

"Jais has defined what an Arabic large language model can be since the first release, and Jais 3 is the clearest step up yet," said Ashish Koshy, CEO of Inception42. "Building an Arabic LLM is not a translation exercise. It means training on the dialects people actually speak, the way they write, and the culture the language carries. That is the work we and IFM have done together across every generation of Jais, and it is why, by bringing together 6 new models, Jais 3 now leads on Arabic across every model size we have built."

Jais 3 is the first adaptation of K2 Horizon, the fleet of six open models IFM released in September, to a new language. The six Jais 3 language models range from a 900-million-parameter model small enough to run on a phone to a 375-billion-parameter flagship. All share the same training recipe and agent capabilities, as well as context window of 512,000 tokens, the longest of any Arabic-native model. Developers can choose the size that fits their hardware and budget. At 7 billion parameters, the average score on IFM's and Inception42's Arabic evaluations rises from 56.4 for the K2 Horizon base model to 79.6 for Jais 3, showing that an open foundation can carry over to a new language and culture without starting from scratch.

Alongside the models, IFM is releasing the audited Arabic benchmarks used in its evaluation and the code to run them. Each model also ships with five training checkpoints, letting researchers trace its development and build from any stage. Enterprises and governments can adapt the models and decide where to deploy them, in line with sovereign AI requirements.

Jais 3 was created by the Institute of Foundation Models (IFM) at MBZUAI and Inception42, in collaboration with Cerebras Systems.

Jais 3 and Jais Speech are available today on Hugging Face, with deployment support for vLLM and SGLang. Learn more at https://ifm.ai/jais/

About the Institute of Foundation Models

The Institute of Foundation Models (IFM) is a global AI research lab dedicated to open and independent development of frontier-class foundation models. Launched in May 2025 by MBZUAI, IFM operates from advanced lab facilities in Abu Dhabi, Silicon Valley and Paris. IFM's model portfolio includes K2 series, the fully open-source fleet of LLMs; Jais, the most advanced Arabic LLM; and PAN, the next-gen World Model for embodied reasoning and real-world simulation.

About Mohamed bin Zayed University of Artificial Intelligence

Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) is a research university in Abu Dhabi. Founded in 2019, it enrolled its first graduate students in 2021 and opened a highly selective undergraduate program in 2025. Recognized globally for its research excellence, it has nearly a thousand students from more than 70 countries. Learn more at mbzuai.ac.ae.

About Inception42

Inception42, G42's sovereign agentic AI company, transforms world-leading research into live deployments at the core of governments and businesses through a new generation of AI products and solutions. Together, they form an intelligence fabric that enables leaders to operate with certainty, act decisively, and deliver real-world impact at national and enterprise scale. AI-native. Human-led

About Cerebras Systems 

Cerebras Systems (NASDAQ: CBRS) builds the world's fastest AI infrastructure. The Cerebras team of pioneering computer architects, computer scientists, AI researchers, and engineers of all types came together to make AI blisteringly fast through innovation and invention. We believe that when AI is fast, it will change the world. Leading global corporations, research institutes, and governments choose Cerebras to run their AI workloads. Cerebras solutions are available on premises and in the cloud. Visit cerebras.ai for more. 

View original content to download multimedia:https://www.prnewswire.com/news-releases/institute-of-foundation-models-and-inception42-launch-jais-3-the-worlds-leading-arabic-native-ai-for-speech-writing-and-agents-302909754.html

SOURCE Institute of Foundation Models