Tired of ads? Enjoy an ad-free experience by signing up.
👩‍🍳 How we use AI at Tech in Asia, thoughtfully and responsibly.
🧔‍♂️ A friendly human may check it before it goes live. More news here

Abu Dhabi launches low-cost AI model to rival OpenAI, DeepSeek

The Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) in the United Arab Emirates has released a new AI model, K2 Think, designed to compete with systems from OpenAI and DeepSeek.

MBZUAI, a research university established by the United Arab Emirates, developed K2 Think in partnership with local AI firm G42, building on Alibaba’s open-source Qwen 2.5 model, and used hardware from US chipmaker Cerebras for testing.

K2 Think has 32 billion parameters, significantly fewer than DeepSeek’s R1 model, which has 671 billion, and is smaller than similar models from OpenAI, which does not disclose parameter counts.

The researchers say K2 Think performs comparably to flagship models from OpenAI and DeepSeek on benchmarks related to math, coding, and science.

K2 Think is intended for specialized applications in math and science rather than general-purpose chatbots.

🔗 Source: CNBC

🧠 Food for thought

Implications, context, and why it matters.

Gulf states are leveraging AI development to reduce oil dependency and gain geopolitical influence

  • The UAE’s K2 Think release represents part of a broader $500 billion Stargate Project initiative to create an AI campus in Abu Dhabi, involving major firms like OpenAI and Nvidia1.
  • This strategic shift reflects what experts call a move from “compute, not crude” as the foundation for U.S.-Gulf relations, with over $200 billion in commercial deals now focusing on technology and AI investments rather than traditional oil partnerships1.
  • The UAE has positioned itself to offer AI services to emerging markets, potentially becoming a regional hub that bridges Western AI capabilities with developing economies that lack the capital and infrastructure of U.S. firms1.

Smaller, specialized AI models challenge the “bigger is better” paradigm in AI development

  • K2 Think’s performance demonstrates a significant efficiency breakthrough, achieving results comparable to much larger models while using just 32 billion parameters compared to DeepSeek R1’s 671 billion parameters.
  • This efficiency gain comes through specialized techniques like long chain-of-thought supervised fine-tuning and test-time scaling, suggesting that targeted optimization can rival brute-force scaling approaches.
  • The model’s focus on scientific applications rather than general chatbot functionality reflects a trend toward purpose-built AI systems that excel in specific domains rather than attempting broad generalization.

Stay ahead in Asia’s tech landscape

You've reached your 2 free content limit for the month. Sign up for free to read the full story.

🏄 For casual readers / 👶 Free

Basic

US$0

Free forever

Get instant access to this article and more every month

0 premium content

Unlimited news briefs

5

5 articles

Ad-free reading experience

Just US$0 per day

⌛Sign up in 20s. No payment details needed.

📖 For learners / 👍 Starter

Lite

US$4.92/month

Billed annually at US$59/year

Get instant access to this article and more every month

4

4 premium content

Unlimited news briefs & articles

Ad-free reading experience

Just US$0.17 per day

Cancel anytime

Our subscriber community includes professionals from these companies:

Stay updated on the go with our mobile app.

Get latest insights with smoother, more personalized experience through TIA mobile app.