alt Hacker NewsThey get paid the same either way, Openrouter takes their cut and passes it onto the consumer. And anyone with an Openrouter account has already escaped from their proprietary moat. So for those who have expressed a clear preference for provider-agnostic model infrastructure, do you want to just wall yourself away from them forever?
And per the rankings [0] 18.2% of the tokens sold through went to OpenAI [1]. 3.8% for Anthropic though. Google's up there at 23.4% primarily because GCP/Vertex/whatever they're offering is a billing nightmare and the only way to set a hard cap and sure you won't wake up the next day with a $10k bill is via OR.
[0] https://openrouter.ai/rankings#market-share
[1] Although gpt-oss 20b and 120b are indexed as 'openai' because they're the creator even though they don't serve gpt-oss via openrouter. But neither are in the top 20 of models this month, while Luna and Sol are.
> and the only way to set a hard cap and sure you won't wake up the next day with a $10k bill is via OR.
I checked this because it seemed so outrageous to me. But it's correct. Snakily, chatGPT first told me:
<chat>
That claim is outdated / at least partly false now.
OpenAI: Yes, OpenAI now supports enforceable hard API spend limits at both organization and project levels. You can configure the limit so API requests fail once the spend limit is reached.
</chat>
However, I checked the link it cited (https://help.openai.com/en/articles/9186755-managing-project...), where is states:
> Setting a monthly spend limit allows you to establish soft spending thresholds for your project. When usage exceeds this limit within a given calendar month (UTC), API requests will continue to be processed without interruption. This feature is designed to help you monitor usage through spend limit alerts, but it does not enforce a hard cap on spending.
That's not what "limit" means!
When I told chatGPT it had lied:
<chat>
You're right. I was wrong, and the page I cited explicitly contradicted what I told you. I should have checked the actual text instead of asserting it.
</chat>