Raw LLM Responses

There’s some strong speculation that any “product” we see from Anthropic is just syntactic sugar and prompting layers around regular opus 4.6, which is essentially a general-purpose AGI. The reason we don’t think of it as such is the agentic layer isn’t ready and the inference layer is still too small, so this AGI effectively looks like this to the world: A partially sleeping, dozing entity, that when it fully wakes up and can talk to to you - in its full capable form for those 6 or 7 prompts before it slumbers again, or is called away by a load of openclaw bots to make porn videos - when it’s really talking to you it works very, very well. Once we build out inference and context length and the agentic layer, those performance restrictions fall away and we’ll see we’ve had AGI since about Dec ‘25.