Cursor has changed a lot in a short time. On October 12, it moved its cheapest plan to Grok-only models, and on November 12, OpenAI will stop providing its models to Cursor. Teams that chose Cursor for access to the best models now have reason to ask whether that access will last.
One of Kilo’s primary architectural principles is model freedom; we believe the best developer tools let you use the best model for the task, not just the models that a certain provider is incentivized to serve.
If your company pays for Cursor and you heard the recent news, you can run Kilo Enterprise free for the rest of that contract, for up to one year. Claim it on the Kilo offer page before the end of November: Get Free Kilo
What changed at Cursor
Two recent changes matter for teams planning their next year:
- October 12: Cursor launched the X Subscription. Its $8 Plus plan now includes Grok models only. The Premium, Super and Ultra plans include no allowance for third-party models, so a request to GPT-6, for example, bills at API list price from the first token.
- November 12: OpenAI ends Cursor’s access to its models and won’t provide future models to Cursor. After that date, a personal OpenAI API key works only for local Chat and Agent.
Model freedom
A model dropdown is not necessarily model freedom. A dropdown shows the models your vendor has a deal with this quarter, and when those deals change, your team’s workflow changes with them. Cursor customers are seeing that happen twice in one month.
Open model freedom means your team keeps the choice. Kilo is open source, so the agent itself isn’t tied to any one model company. The Kilo Gateway lists 500+ closed and open-weight models at the provider’s price, so a team can run GPT for one task and an open model for the next.
Our own usage data shows why this is valuable - when developers balance cost and performance on their own, they use 3+ model providers up to 70% of the time. And with Kilo enterprise controls, they can make those decisions safely, and within your org’s compliance.
We made the full case in Your coding tool should not choose.
Keep your GPT models after November 12
On September 29, we launched Sign in with ChatGPT. Teams can connect their ChatGPT plan to Kilo and use it in VS Code, JetBrains, the CLI, Cloud Agents, Code Reviews, and the mobile app. Requests draw on the ChatGPT Work and Codex usage included in the plan, up to 100% or a lower limit you set in ChatGPT, instead of per-token API billing.
For Cursor teams facing the OpenAI cutoff, this means GPT access runs through the plan they already pay for. The same dropdown lists the 500+ Gateway models beside it, so nobody has to choose between OpenAI and everything else. OpenAI models have been at the top of the Kilo Leaderboard since we launched it.
What Kilo gives enterprise teams
Kilo is built so no single vendor deal decides which models your engineers can use.
- 500+ models at the provider’s price. Each session shows its cost as it runs, so nobody finds out at the end of the month.
- Admin controls per sub-org. Admins set model and provider rules for each sub-org from one screen.
- Auto-Efficient routing. Kilo sends each prompt to the cheapest model proven capable of the task.
Anaconda’s own engineers made this move. Read what they found when they moved from Cursor to Kilo, or compare the two products feature by feature in Kilo Code vs Cursor. If you’re weighing other options too, see our roundup of open-source Cursor alternatives.
How to switch
- Submit the claim form with proof of your Cursor contract on the offer page. The page also has a short FAQ on eligibility and verification.
- Follow the Cursor to Kilo migration guide to move your team over.
- Set your model and provider rules per sub-org, then turn on Auto-Efficient routing.
If November 12 is on your calendar, switching now gives your team a month to settle in before the OpenAI cutoff: Get Free Kilo