DeepSeek

DeepSeek released R1 in January 2025 as an open-weight reasoning model that matched or beat several closed frontier models on math and coding benchmarks — at a fraction of the training and inference cost — and the announcement briefly wiped hundreds of billions off Nvidia's market cap as investors reassessed how much compute frontier AI actually requires. Because the weights are open, researchers and companies can self-host, fine-tune, or build on it instead of depending on a single vendor's API, which is exactly what a lot of the industry has since done.