@888Digital888 I think you're right! High reasoning effort just means more tokens are used. For some reason I thought it's different versions of the model. They should've named reasoning_effort as thinking_time or thinking_token_usage.

@marc_louvion If you're a developer, learn how to change your identity from a developer to a founder/entrepreneur. This will be the most impactful work you can do on yourself. Everything else follows.

@ericzakariasson @cursor_ai I launch new composers mid-task if convos get long. Years of working with OpenAI models from the time they came out taught me to think about context like this. Would be nice to see our context usage % now that we have project rules, docs and custom prompts from cursor side.

@AravSrinivas What query was that? I’ve never seen it going that deep and I’d like to try

@dannypostmaa I kinda see cursor as my teammate. And that team member can use Claude, Deepseek or whatever model is the best to get the job done. Application layer wins imo.

@houman_X @dannypostmaa Don't think that's the case. Data is sent to Anthropic.

@AravSrinivas How would this work? On Perplexity mac app? What would be the minimum memory required?

@brian_lovin This is the way. Also make sure you’re using the agent mode in composer.

@ozenhati @GroqInc Thank you 🙏

@iannuttall Cursor has a generate commit message button btw

I think people aren't loyal to big platforms. They'll go where they can sell for more without getting scammed. The real challenge? Getting traffic to your new site. Fastest validation: reach out to active auctions on acquire/flippa, make sites for each, send to your email list of optic auction subscribers. Test if people will bid/buy.

@ozenhati @GroqInc Correct me if I'm wrong, but setting a budget is only available with LiteLLM Proxy, so it's a bit cumbersome. Please prioritize a spend limit or just let us add money like on OpenAI and Anthropic.

@simonw @karpathy @levelsio I'm using it for real-time dictation, and it works great. You can set a key (I'm using the right option key) for push to talk too.

@gregisenberg And the cost is dropping fast in AI. Deepseek R1 and Gemini 2.0 Flash came out 2 weeks apart and prices are way down.

@ShaanVP You can just do things

@p_millerd To try something different: - Add wordpress docs from cursor settings - Give the wordpress docs as context to agent

@p733dev @levelsio @karpathy It's mentioned in the gumroad page. The pro features are under MacWhisper Pro heading