Skip to content

EthenEthenEthen

LoginPlatformTry Ethen

Open Source Model Profile · QCRI

Fanar-1-9B-Instruct

Fanar-1-9B-Instruct is an 8.78B-parameter Gemma 2 Arabic-English instruction-tuned fine-tune from QCRI. According to the model card, it continues google/gemma-2-9b on 1T tokens with SFT and DPO post-training.

Publisher
QCRI
Task
text-generation
Model type
gemma2
License
apache-2.0
Library
transformers
Publication status
Accepted · not indexed

Model overview

Fanar-1-9B-Instruct is published by QCRI as a gemma2-based text-generation model. The captured configuration identifies Gemma2ForCausalLM, and Safetensors metadata reports 8783871488 parameters. According to the model card, it is the instruction-tuned version of Fanar-1-9B, continually pretrained from google/gemma-2-9b on 1T Arabic and English tokens with 4.5M SFT instructions and 250K DPO pairs.

Recorded capabilities

8.78B Gemma 2 scale

Captured configuration identifies Gemma2ForCausalLM with a gemma2 model type, and Safetensors metadata reports 8783871488 parameters.

Arabic-English instruction tuning

According to the model card, it is the instruction-tuned version of Fanar-1-9B, continued from google/gemma-2-9b with attention to Modern Standard Arabic and Gulf, Levantine, and Egyptian dialects.

Documented 1T-token and SFT plus DPO recipe

The model card describes continual pretraining on 1T tokens, then SFT on 4.5M instructions and DPO on 250K preference pairs with the LitGPT framework.

Use cases in the source record

  • Arabic-English conversational text generation, including Modern Standard Arabic and the Gulf, Levantine, and Egyptian dialects described in the card.
  • Transformers and vLLM chat-template inference experiments using the card's documented loading and generation workflow.

Limitations and unknowns

  • No evaluation results were extracted from this record.
  • The model card states a 4096-token context length; no independent context measurement was extracted.
  • Provider state is historical snapshot data, not independently refreshed current availability.
  • Broader Fanar GenAI platform capabilities described in the card, such as image generation and TTS or ASR, are platform claims and were not extracted as capabilities of this text-only model.

Source and provenance

Source: QCRI/Fanar-1-9B-Instruct

Captured: Unknown. Processed: 2026-09-07T19:35:51.158709+00:00.

Fanar-1-9B-Instruct Fanar-1-9B-Instruct is a powerful Arabic-English LLM developed by Qatar Computing Research Institute (QCRI) at Hamad Bin Khalifa University (HBKU) , a member of Qatar Foundation for Education, Science, and Community Development. It is the instruction-tuned version of Fanar-1-9B . We continually pretrain the google/gemma-2-9b model on 1T Arabic and English tokens. We pay particular attention to the richness of the Arabic language by supporting Modern Standard Arabic (MSA) and a diverse set of Arabic dialects, including Gulf, Levantine, and Egyptian. Fanar models, through meticulous curation of the pretraining and…

F001F002F003F004F005F006F007F009F010F011F012F014F015F016F017F018