Reasoning models have shown progress on math, code, and science benchmarks.OpenThoughts project aims to create open-source datasets for training reasoning models.OpenThoughts2-1M dataset led to OpenThinker2-32B, the first model trained on public reasoning data to match DeepSeek-R1-Distill-32B.OpenThinker3-7B model achieved state-of-the-art results on benchmarks like AIME 2025 and GPQA Diamond.