KSA, Riyadh — WideBot AI, the foremost provider of conversational AI in the MENA region, proudly announces the launch of AQL-7B, a revolutionary Arabic Large Language Model (LLM) that has achieved the highest evaluation results in Arabic AI performance benchmarks. AQL-7B leads in accuracy and positions itself as a superior model for real-world applications, offering both efficiency and flexibility. Additionally, AQL-7B's compact model size (7 billion parameters) gives it a competitive edge over all Arabic large language models, making it more efficient while maintaining high performance.

“With AQL-7B LLM, we’ve surpassed all existing Arabic LLMs, pushing the boundaries of what’s possible in general Arabic AI,” said Mohamed Mostafa, CTO of WideBot AI. “AQL-7B demonstrates our AI lab's commitment to excellence, and we’re excited about its potential to transform how businesses in the MENA region engage with AI.”

Key Features of AQL-7B:

  • Unmatched Performance: AQL-7B outshines other Arabic AI models in various leaderboard benchmark datasets such as MMLU, ARC Challenge, PIQA, and RACE. Despite having only 7 billion parameters, AQL-7B competes directly with models over 10 times its size, proving its efficiency and prowess.
  • Compact & Efficient: Designed for speed and cost-efficiency, AQL-7B delivers high performance without requiring expensive infrastructure, making advanced AI accessible to a wide range of users and businesses.
  • Multilingual Capabilities: Beyond Modern Standard Arabic (MSA) and dialects, AQL-7B supports over 30 languages, including:
    • Middle East: Persian, Hebrew, Turkish.
    • Western Europe: German, French, Spanish, Portuguese, Italian, Dutch.
    • Eastern & Central Europe: Russian, Czech, Polish.
    • Southern Asia: Hindi, Bengali, Urdu.
    • Eastern Asia: Japanese, Korean.
    • South-Eastern Asia: Vietnamese, Thai, Indonesian, Malay, Lao, Burmese, Cebuano, Khmer, Tagalog.

Superior Model Evaluation: In rigorous testing against other prominent LLMs, AQL-7B excelled across key performance indicators, demonstrating leading accuracy and consistency.

Dataset

AQL-7B

SILMA-9B

Jais-Inception-70B

mistral-nemo-wissenschaft-12B

calme-2.2-llama3-70b Llama-3.1-70B

CohereForAI c4ai-command-r-plus-08-2024

AceGPT-v2-32B

Microsoft Phi-3.5-mini-instruct

ACVA

81.22

78.89

54.11

40.14

53.31

39.79

39.51

66.39

AlGhafa

73.4

71.85

57.85

49.62

56.63

53.83

54.93

47.69

MMLU

55.18

52.55

62.45

42.56

58

57.5

46.2

37.74

EXAMS

52.93

51.4

55.49

44.51

56.24

53.07

47.11

37.99

ARC Challenge

79.65

78.19

59.14

43.45

54.66

57.84

49.83

58.36

ARC Easy

87.31

86

60.53

45.3

56.94

60.7

54.78

37.99

BOOLQ

66.79

64.05

80.21

81.41

84.57

85.89

62.12

70.12

COPA

80.05

78.89

57.78

55.56

66.67

48.89

46.67

53.33

HELLASWAG

50.42

47.64

38.59

28.83

47.92

35.29

33.66

43.87

OpenBookQA

75.11

72.93

57.98

48.28

54.55

53.74

51.72

55.35

PIQA

74.28

71.96

73.43

60.88

69.89

68.69

65.79

70.21

RACE

76.89

75.55

56.95

45.04

55.39

57.33

48.55

58.94

SCIQ

92.84

91.26

64.52

56.58

70.05

50.25

67.14

66.63

TOXIGEN

70.07

67.59

57.97

62.46

72.51

56.79

43.21

67.81

Average

72.58

70.63

59.79

50.33

61.24

55.69

50.8

55.17

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

AQL-7B Model evaluation based on Arabic LLMs leaderboard datasets:

AQL-7B is part of WideBot AI’s ecosystem and is available across multiple channels, including Messenger, WhatsApp, Instagram, Twitter, WebChat, and external APIs providing businesses with the tools to implement AI across various platforms and languages.

Empowering Businesses in the MENA Region:

AQL-7B is set to drive innovation in key sectors such as Governments, e-commerce, banking, healthcare, telecommunication companies, and more, providing tailored, localized solutions in Arabic and other languages. By integrating AQL-7B into their systems, businesses can harness advanced AI capabilities to improve customer experiences, optimize workflows, and enhance decision-making processes.

“AQL-7B is not just another LLM—it’s a leap forward for AI in the Arabic-speaking world. It will unlock new opportunities for businesses to create smarter, more intuitive AI-driven experiences,” said Nabil, CEO of WideBot AI.

WideBot AI's Vision for AI Solutions:

WideBot AI continues its commitment to developing AI models that are not only powerful but also responsible. AQL-7B has built-in security measures and our services are ISO27001 certified to ensure its ethical use across various applications. WideBot AI aims to bring AQL services to over 30% of companies and governmental entities in the Middle East and North Africa by 2026.

About WideBot AI:

Founded in 2016, WideBot AI has consistently led the Arabic AI landscape by offering cutting-edge conversational AI solutions that cater to the unique needs of the MENA region. With a focus on Arabic dialects and advanced AI technology, WideBot empowers enterprises with innovative tools to engage with their customers effectively.For more information on AQL-7B, visit www.widebot.ai experience WideBot platform, and ask your most challenging questions.

Contact Information:
WideBot AI
Website: https://widebot.ai
Email: aql@widebot.ai

#AI #ArabicLLM #WideBot #AQL #NLP #ArtificialIntelligence #MachineLearning #AIInnovation #MENA #Tech #DataScience #AIModels #NLP