Next-Generation Arabic LLM: Introducing a significantly larger Large Language Model (LLM) for unparalleled Arabic fluency and content generation, increasing the model size from 9 billion to 27 billion parameters.
Context & Memory Boost: Expanded context size allowing the LLM to process and retain a richer context, leading to smarter, more relevant, and longer conversations and document analysis.
Bilingual Reasoning: Moving beyond simple conversation, integrating inherent reasoning capabilities allows the LLM to perform strategic thinking and complex decision-making, while providing a step-by-step explanation of its thought process, delivered in Arabic.
Arabic Poetry Generation: A new dedicated model focused specifically on generating high-quality, authentic Arabic poetry.
Advanced Prompting (Multi-Action): Support for complex multi-step instructions using built-in tool-calling capabilities, allowing Fanar to execute several tasks from a single prompt, for example: “translate this article, summarize it, and create a social post based on the summary”.
Dialect-Aware Speech-to-Text: A new model that accurately transcribes longer audio clips while recognizing the various speakers—perfect for conversation summaries and meeting analysis.
Expanded Text-to-Speech (TTS): A more natural and expressive TTS model with eight new voices in both Arabic and English, including new voice-personalization capability.
Culturally Intelligent Image & Video AI: Upgraded image and video understanding and generation models that are natively fluent in both Arabic and English, and specifically trained to be culturally aware and appropriate.
Comprehensive Content Moderation: A new culturally aware filtering model that works tightly with all Fanar’s text and image generation models to accurately flag and prevent the creation of harmful or inappropriate content.
Fresh User Interface: A brand-new look for both our web chat and mobile applications, providing a seamless and intuitive user experience.