RE: DeepSeek R1 moment, open-source now competing with the best OpenAI reasoning models, you can even watch DeepSeek thinking out loud. Full transcript (DeepSeek R1 Distill Llama 70B).

[vikisecrets]

Yes, it's more or less that useful as o1 or GPT 4o, nice to play with it, but not super useful in a commercial sense. Struggles with more complex tasks like all GPT models. Even has problems writing an error-free hello world program as shown above.