2025 in artificial intelligence

Events

January

• January 20 – DeepSeek releases DeepSeek-R1, a large language model based on DeepSeek-V3 utilising a chain-of-thought, stating it achieves performance comparable to OpenAI-o1 across math, code, and reasoning tasks. DeepSeek-R1 is open-source.
• January 21 – The Stargate Project, a joint venture created by OpenAI, SoftBank, Oracle and MGX, is formally announced by U.S. president Donald Trump.
• January 23 – Humanity's Last Exam, a benchmark for large language models, is published. The dataset consists of 3,000 challenging questions across over a hundred subjects.
• January 27
• * Nvidia's stock falls by as much as 17–18%, after the release of DeepSeek-R1.
• * DeepSeek-R1 surpasses ChatGPT as the most-downloaded free app on the iOS App Store in the United States.

February

• February 2 – Andrej Karpathy coined the term vibe coding to describe how AI tools allow hobbyists to construct apps and websites, just by typing prompts.
• February 3 – OpenAI releases ChatGPT Deep Research, an artificial intelligence system integrated into ChatGPT, which generates cited reports on a user-specified topic by autonomously browsing the web for 5 to 30 minutes.
• February 6 – Mistral AI releases Le Chat, an AI assistant able to answer up to 1,000 words per second.
• February 10
• * AI Action Summit takes place in Paris, France, for two days.
• * It is announced that France will receive 109 billion euros in AI private investments over the coming years.
• * Anthropic launches the Anthropic Economic Index, an initiative aimed at "understanding AI's effects on labor markets and the economy over time".
• * Elon Musk and a group of investors led by him offer to buy OpenAI for $97.4 billion.

• February 27 – OpenAI announces a research preview of GPT-4.5, its largest and most advanced AI model to date.

March

• 27 March – Engineers of the Texas A&M University create an "Super-Turing AI" designed to function more similarly to the human brain. Unlike conventional models that isolate tasks and shuttle massive volumes of data between components, it merges key processes, streamlining operations and enhancing efficiency.
• 31 March – OpenAI's GPT-4.5 is reported to be indistinguishable from a human in text conversations.

April

• 16 April – OpenAI announces the launch of two new AI models, o3 and o4-mini.

May

• 14 May – Google DeepMind announces AlphaEvolve, a Gemini-powered coding agent for designing advanced algorithms.
• 20 May
• * Google launches A.I. Mode, which will be a feature on their search engine, and uses the Gemini model.
• * Google DeepMind announces Veo 3, a new state-of-the-art video generation model. The company also boosts the performance of Gemini 2.5 Pro, its flagship AI model.
• 22 May – Anthropic releases Claude 4, with two models: Claude Opus 4 and Claude Sonnet 4. According to Anthropic, Claude 4 can function on its own for hours.

August

• 7 August – OpenAI releases GPT-5.

September

• 8 September – Artificial intelligence uses vast energy, but University of Florida researchers develops a chip that replaces electricity with light for key AI tasks. Using microscopic lenses etched onto silicon, it performs laser-powered computations with drastically lower energy and near-perfect accuracy.

November

• The Mission Genesis launch