Tweet by fnthawar

June 4, 2025

I'm seeing a concerning trend in talking to CTOs/CEOs about the cost of tokens i.e. Can I afford an extra $1-10k/month per engineer as they use @cursor_ai / @windsurf_ai / @GitHubCopilot etc? So they clamp down on spending 😬 1. Do not use the default / cheapest models. Yes all the models are getting better, but there is a big difference between the default Cursor model, the small @GoogleDeepMind Gemini Flash and @OpenAI O1 Pro. Splurge for a better model. 2. You might use completely different models for planning and actual coding. Exampe — O1 pro for planning / requirements and then @AnthropicAI Claude 4.0 for coding 3. If your engineers are spending $1k/month MORE because of LLMs and they are 10% more productive, then that is TOO CHEAP. i.e. Anyone would KILL for a 10% increase in productivity for only $1k/month. Totally worth it 4. If you are worried about engineers spending $10k/month on tokens -> I wouldn't worry. If they are spending this much, DM me because I want to know what exciting things they have conjured up to spin up LLMs on. If they are spending this much AND getting value out of it, you are winning (DM me!) 5. Short of tempting Goodharts Law, you could reward your engineers (well, all employees!) for maximal AI spending -> this means they have figured out how to leverage the LLM well 6. The workflow is moving quickly from AI autocomplete to "I am the architect of a team of agents". This means a single person can wield of a team of LLMs to build their code. Makes sense to see your corresponding token bill go 📈 7. Worry about LOC (lines of code), as the current cohort of models is very code verbose. This will hurt you later if you're building for production vs. just prototyping. It will take smart prompting / rules / iterations to build the short / sweet elegant code you actually want 8. My Cursor deployment quickly increased when non-RnD folks started using it to build their workflows. Folks in Support, Commercial, Ops and more can build things now 👊🏽. The last thing I want is to stop these licenses from growing. I know folks @ShopifyEng who were proud to show up on the Top 10 token spend leaderboard due to the valuable work being done

Author
fnthawar
Date
June 4, 2025
Open on X