OpenAI just dropped the biggest base model on the internet

OpenAI published the weights for o1, its most capable reasoning model. The model is 1.4 trillion parameters and 256K context. It scores 90% on MATH and 92% on AIME 2024 — both benchmarks that previously only top research labs could touch. The weights are on Hugging Face, so anyone can run them, fine-tune them, or build on top.

The important part is what this means for the open weights race. Open weights let the community fork models, strip away the safety wrappers, and push them into places the original companies never intended. It also means the open source ecosystem gets access to a model that rivals the closed ones — without paying the API bill.

The broader pattern is shifting: base models used to be closed because they were too expensive to run. Now the big labs are publishing them anyway, partly for visibility, partly because the market is moving faster than their own distribution channels. The 1.4T size means this model needs real compute to run — a single A100 won't cut it — but the weights themselves change the landscape.

Why this matters for us: it means the next generation of tools we build for our communities can run on models we actually control, not just on APIs where someone can pull the plug or raise prices overnight.