Hey guys, I was just thinking about this idea and wanted to share it with you all.
Since Cursor released Composer-1, there has been less visibility into the fact that such a model exists, because it looks like Cursor spends little on marketing (YouTube, Facebook, etc.). Also, this model, built with data from their own users, is sold back to those users on a cost-per-use basis, which is absurd in my opinion. If you want to take market share from your competitors, you’ve got to give users an incentive to use your model over others, and then monetise later. Your first job should be to grab market share. People should feel like they are paying for something meaningful.
Hence, I propose giving Composer-1 away for free under capacity limits (3 million tokens per week or something).
For example, grok-code-fast-1 is offered for free, so users use it even if they do not like it, because they are incentivised to do so. Therefore, Grok code is always on a Cursor user’s radar, as they are almost certainly going to run out of credits for premium models. This keeps grok-code-fast-1 in the game due to demand, even though they are not winning, rather than being lost in the shadows like Deepseek, kimi k2, and older GPT models in Cursor.
Let’s say you provide Composer-1 for free under capacity limits. What would this translate to?
- You get a bunch of more users on your platform
- DAP increases, hence your shareholder meetings will love you - raising more $$$ rounds is how you monetarily win here, this pays for your model inference costs
- You can provide higher capacity limits for the free Composer-1 when the user purchases a paid subscription. For example, 3 million free tokens for free users, 6 million free tokens for paying users, which is +100% usage for free
- Conversions increase, as people are actually paying you for using your model (for free, wouldn’t this be insane?). For example, 100 out of 1000 free users sign up - that’s $2000 cashflow per month - this pays for model development costs
- CC users who left Cursor might come back as Anthropic is cash-cow-ing their users and will never do this; everybody loves to save cash
Well, that’s just an idea I had; you might want to run it by your product and finance teams to see if the model could work. Like always, feel free share what you think about this.