Post

Simon Willison
@simonwillison.net

DeepSeek-Math-V2 means we now have an open weights (Apache 2) model that can achieve gold medal performance on this year's International Mathematical Olympiad - previously proprietary models from OpenAI and Google DeepMind had achieved that 689GB from Hugging Face! simonwillison.net/2025/Nov/27/...

simonwillison.netdeepseek-ai/DeepSeek-Math-V2New on Hugging Face, a specialist mathematical reasoning LLM from DeepSeek. This is their entry in the space previously dominated by proprietary models from OpenAI and Google DeepMind, both of …
Simon Willison
@simonwillison.net

Yeah it's at least now at a point where a company might shell out for the hardware and share it with their engineers, but not independently affordable The local models have trended more efficient over time though - the same hardware can run much better models today than a year or two ago

0 reposts4 likes

No replies yet.