i think we have the causation backwards here. llms aren't expensive because they have to be — they're expensive because we keep reaching for the expensive model instead of putting any effort into making the cheap one good enough.
a surprisingly large fraction of production workloads can be handled by smaller models with the right scaffolding. it's often easier to switch to a larger model than to engineer those pieces, so many teams never bother.
my intuition is that a lot of the current "ai cost crisis" is really an orchestration problem rather than a model pricing problem. before asking whether frontier pricing is sustainable, i'd first ask how much of that spend is simple tasks being sent to the smartest available model by default.
my bet for the next few years is that the model itself stops being where the value is. frontier models will become more like commodities, and the real difference will be the layer around them as routing each task to the cheapest model that can do it well, verifying the output, and only escalating when needed.
eventually, asking "which model do you use?" will sound a bit like asking "which cpu do you use?" the engine still matters, but the system built around it matters a lot more.
what llms reason best in might not be a language at all. more like a graph. graphs say more in less space than prose, and they stop the model from wandering off.
not the author of the gea but from what I can see in the readme, the ideas go back to 2017, erste.js and regie were earlier versions of the same concept.
not the author of the gea but from what I can see in the readme, the ideas go back to 2017, erste.js and regie were earlier versions of the same concept.
For a fun side project, we built a quick demo to interview our OpenClaw agent called Mahmut. We used LemonSlice, LiveKit, and ElevenLabs. We just shared the full interview, and here's the repo if anyone is interested in building the same thing: https://github.com/openserv-labs/openclaw-voice-avatar
Our avatar was a really hard one to handle, but I think it's still really good. All we did was connect the APIs. LemonSlice is really cool.
I just found about them on twitter and tested, its so crazy that it work so smooth and fast. It's probably powered by Claude Agent SDK in behind, I created a fully functional telegram bot under 3 minutes. It seems perfect for non-technical users.
reply