AI compute shortage isnโt going to be solved anytime soon. But, modelโs efficiency can.
Modelโs ability to draw less compute, caching, less memory throughput, efficient agent tool calls is going to make a lot of traction.
#prediction
@karpathy Managing the AI generated tools or mission control dashboard for Claws seems still unorganised. I would imagine it is going to be a thing in very near future. :)
@karpathy Managing the AI generated tools or mission control dashboard for Claws seems still unorganised. I would imagine it is going to be a thing in very near future. :)
Gemini 3.1 Pro is now available for all paid tiers!
The default model router, Auto (Gemini 3) will use Gemini 3.1 Pro as it's pro model for complex prompts.
You can also set the new model via /model to try it out.
We are excited to see how you put it to use!
Finding a commercial space in Whitefield, Bangalore is like finding a shade in a desert. The concrete jungle has left no space for small businesses to start a new.
Instead of building yet another SaaS product, build products that help busy or non-technical people create their own solutions for their specific niche problems.
Claude 20$/mo subscription gives you only a preview of what Opus 4.5 model could do. Forces you to upgrade to 200$/mo because the current rate limit is so low.
Here is the usage limit with just one or two prompts. Completely unusable!
Prompt UI in Codex app and the Codex VS Code extension UI looks exactly the same. Makes more sense to distribute it as electron app for consistency and reuse.