Our Terms & Conditions | Our Privacy Policy
Even as the Chinese low-cost Large Language Model (LLM) DeepSeek-R1 creates ripples across the LLM community globally, the Indian ecosystem is evaluating the pros and cons of the new approach to LLM as it softens the entry-level barriers.
“It has significantly lowered the cost of LLM inference. The cost of using DeepSeek’s models is about one-tenth of what OpenAI charges per million tokens. This cost reduction is a major factor that allows companies to build applications using DeepSeek and other similar models rather than relying on more expensive options,” Prof. Balaraman Ravindran, the Founder and Head of the Wadhwani School of Data Science and AI, at IIT Madras, told businessline.
This is a significant advantage, especially for smaller companies and start-ups that may not have the resources to afford the costs associated with larger models like those from OpenAI.
- Read: DeepSeek: The AI disruptor you need to know about
API Access
“DeepSeek’s models can be accessed via APIs, similar to how people use GPT4 or Llama. This enables businesses to integrate these models into their applications without the need for substantial investment in infrastructure,” Ravindran, who also heads the Robert Bosch Centre for Data Science & AI and the Centre for Responsible AI (CeRAI), said.
The cost per token for DeepSeek is significantly lower than that of OpenAI, with some models, like the one from ByteDance (owner of TikTok), being even cheaper than DeepSeek. This price difference allows more companies to build and operate applications using these LLMs.
He also notes that building foundation models is now more affordable. He gives the example that DeepSeek’s model cost approximately six million dollars to run, which is significantly less than what OpenAI has spent
With this cost shift, India can become a player in fundamental research in the LLM space. “The cost of entry has been lowered, allowing more groups to engage in research, especially given the government’s plans to invest in GPU infrastructure,” he said.
Y Kiran Chandra, Centre Head of viswam.ai, which is trying to build an LLM for India, said the DeepSeek model was published by DeepSeek under an MIT license, which permits free reuse and modification.
“However, it’s important to note that the model cannot be considered fully open-source because the training data has not been made available. For a project to be truly open-source, all four critical components—algorithms, datasets, model weights, and source code—must be accessible for public review and use,” Kiran, who also heads open-source activists association Swecha, said.
He, however, admits that the release of an ‘open-weight’ model is a significant milestone.
“By making the model weights available, they have enabled researchers and developers to study, adapt, and build upon the algorithm. This approach demonstrates that high-quality AI models can be developed without relying solely on the brute-force, resource-intensive methods often pursued by organisations in the Global North,” he added.
“For India, this presents a unique opportunity. By focusing on creating high-quality, digital-first datasets tailored to local contexts, India can leverage this open-weight model to develop world-class AI systems,” he said.
Wakeup call
Naresh Singh, Senior Director Analyst at Gartner, said the DeepSeek model is a wakeup call for India’s tech sector which “tells them that we must fast track our efforts on the AI front. The world is being divided into the haves and have-nots when it comes to cutting edge AI,” he said.
India, encompassing both the public and private sectors, should urgently focus on identifying the nation’s AI needs and practical use cases to address governance, social, and economic challenges.
“It is essential to develop a robust ecosystem dedicated to building AI models and applications across various sectors, benefiting not only India but also the global community,” he said.
Images are for reference only.Images and contents gathered automatic from google or 3rd party sources.All rights on the images and contents are with their legal original owners.
Aggregated From –
Comments are closed.