Technology
OpenAI with New Models That Mimic Human Reasoning
Apr
OpenAI has unveiled two advanced artificial intelligence models, o3 and o4-mini, designed to simulate human reasoning. This development aims to strengthen its position amid rising competition from the United States and China. Both models can handle visual tasks and complex programming challenges.
Multistep Reasoning and Visual Analysis Models
The o3 model spends more time processing to solve multistep problems in math, science, and coding. Its counterpart, o4-mini, offers a faster and more efficient alternative with impressive performance. Both are available to paid ChatGPT subscribers with full access to its tools.
Visual and Logical Integration: A Milestone in AI
For the first time, these models integrate visual data into the logical process. This enables them to analyze blurry images, rotate or zoom in on photos, and perform complex tasks with greater accuracy. They can also use tools like web browsing and image generation, expanding their versatility across various fields.
Fierce Competition Drives Innovation
Since its breakthrough with GPT-3.5 in 2022, OpenAI has rapidly evolved. However, it now faces challenges from players like DeepSeek (China), Anthropic, and Elon Musk’s company xAI. All have launched models with competitive capabilities, putting pressure on OpenAI’s leadership.
Powell Reinforces Fed’s Stance Amid Inflation Risks from Tariffs
Federal Reserve Chair Jerome Powell reaffirmed on Wednesday the central bank’s commitment to preventing inflation…
Programming: AI’s New Battleground
Programming has become a key sector for generative artificial intelligence. Companies like Google (Alphabet) and startups such as Anysphere, creator of Cursor, are competing fiercely in this space with solutions tailored for developers and coders.
OpenAI Launches Codex CLI for Developers
In response to this trend, OpenAI introduced Codex CLI, an open-source AI agent that runs locally in terminals. Designed to work with models like o3, it aims to assist with coding tasks and promote open usage within the developer community.
Upcoming Releases: Open Model and GPT-5
Sam Altman, CEO of OpenAI, revealed that the company will soon launch an open version of its reasoning model, inspired by the success of DeepSeek’s R1 system. He also confirmed that GPT-5, the highly anticipated next-generation model, will be released “within a few months,” marking another milestone in AI development.
