This year, global big tech corporations are actively pursuing a free strategy for their artificial intelligence (AI) models. The competition among major big tech corporations, including OpenAI, for AI models and services is intensifying as they seek to attract users.
According to the IT industry on the 1st, Google announced on the 30th (local time) through X (formerly Twitter) that it will provide 'Gemini 2.5 Pro' for free to all users. However, free users will have limited usage. Google introduced Gemini 2.5 Pro on the 26th, stating it is 'the most intelligent AI model' designed to solve increasingly complex problems. Gemini 2.5 Pro has demonstrated excellent performance in coding, mathematics, and science on the AI performance evaluation platform LM Arena.
Initially, Google provided this model on its developer platform, 'Google AI Studio,' and through the paid service 'Gemini Advanced' to Gemini application users. However, this measure has given more users the opportunity to experience the latest model. The industry considers it unusual for the latest inference model to be made free less than a week after its release.
Elon Musk, the chief executive officer of Tesla, owns xAI, which also started offering the beta version of its latest AI model, 'Groq3,' for free to users without charges about three days after its release in February, saying, 'We will offer Groq3, the smartest AI in the world, for free until our servers melt down.' At the time of its launch, Elon Musk confidently presented the model as 'the smartest AI' to compete with OpenAI. Less than a week after introducing its latest service for free, xAI appears to be swiftly expanding its user base as a latecomer in the AI assistant service.
Microsoft (MS) opened its AI assistant service, 'Copilot,' allowing all users to use the voice conversation mode and the advanced reasoning function 'Think Deeper' for free without limitations in February. In response, OpenAI also began providing the 'Advanced Voice' feature based on 'GPT-4o Mini' to free users on the same day. This feature, which was first introduced in May last year, allows users to interact with ChatGPT without touching their mobile phones. OpenAI has built a revenue model by previously offering advanced features solely to paid users who pay $20 (29,000 won) monthly.
To secure both general consumers and AI developers within its ecosystem, there is an expansion of the so-called 'open source' strategy, providing AI development tools for free. MS introduced a free version of 'GitHub Copilot,' which assists developers' coding at the end of last year, supporting up to 2,000 code generations for free per month. Furthermore, Google announced in February that its AI coding assistant 'Gemini Code Assist' would support up to 180,000 code generation requests monthly. Meta, which plans to release the 'Llama 4' model in the first half of this year, is also expected to intensify its open-source strategy.
Global big tech corporations are transitioning their AI services to free models as competition for AI models intensifies, leading to a user land grab. The characteristics of AI models, particularly those using large language models (LLMs), make economies of scale vital. Enhanced technology can only be demonstrated if a released service continues to attract many users. Therefore, corporations aim to build long-term technological competitiveness focusing on acquiring users and collecting data rather than achieving short-term profits.
Additionally, the rapid decline in expenses associated with inference as AI technology advances is also cited as a contributing factor. The industry reports that the inference costs for LLMs to deliver the same performance are decreasing to one-tenth each year.