Our take on Cursor's pricing changes

There’s been a lot of chatter about Cursor’s pricing changes over the last few days. You might have seen these viral tweets or their blog post titled “Clarifying Our Pricing”

The unfortunate part is every AI-native product will wrestle with usage-based economics.

Here’s our take on the situation -

What Cursor gets right: 

• Pass-through costs ≠ hidden costs - token mark-up is more honest than flat seats when COGS swing with model choice

• User-supplied keys push power (and arbitrage) to devs who care about tuning spend

Where it still needs work: 

• Predictability - users still need a bill they can forecast. Raw token math ≠ budget clarity

• Risk transfer - “bring your own key” moves volatility off cursor’s books and onto customers’. Great for margin, not for trust

• Education gap - prompt → token → $$$ is still hard to grok

Playbook we see working here: 

• Hybrid pricing - low, predictable platform fee plus transparent compute pass-through with caps

• Real-time cost telemetry - show usage → spend in-product (minutes, not months - at Quivly we help our customers with this)

• Guardrails by default - soft + hard limits so users (or enterprise finance teams) are never surprised

If you are looking to give your customers visibility into real-time usage and their expected billing and want to know how we can help, please reach out or find some time here: cal.com/chandrika