> These companies only offer closed versions of their powerful models (despite the word "open" in "OpenAI"). You pay them for chat and/or API access, and they are allowed to look at everything you do.
How should I reconcile this claim with 'Google Vertex' and 'Amazon Bedrock' being offered as providers with zero data retention on https://openrouter.ai/anthropic/claude-opus-5.5#providers?
My guess is the average Joe will get rejected, but if you're a big company who cares about trade secrets, etc - of course the major cloud providers will provide a ZDR option.
EDIT: strangely Azure shows up as ZDR on openai's models but not anthropic's: https://openrouter.ai/openai/gpt-6.1-sol#providers
For clarity, I do not think you should use abusive language with Ai because of the impact it has on you, Ai only simulates emotion through language use, there are no hormones at work.
Not sure what you mean.
By default, Azure retains. Always had. The ZDR is a special status given after review, and likely not available to everyone.
When Anthropic was requiring retention for Fable, our company didn't get access to it via the cloud provider for precisely this reason: The provider said Anthropic was requiring retention. We still had access to Opus, etc.
https://openrouter.ai/docs/guides/features/zdr
They give administrators toggles to turn off models which do not fit company ZDR policies, in part because Anthropic forces resellers to comply with Anthropic's (eg.) 30 day retention.
Their vertex endpoint is changing because it is not zdr.
Fable does not have a zdr endpoint for any company, any platform.
Bedrock of course has their flagged prompts loophole.
You can't trust them, in any case, the endpoint could be anything. You can barely trust TEE, sadly the tinfoil endpoint I used is heading towards surveillance as well by adding abuse monitoring.
Since you seem to know, it's one thing to not retain any data long term, but is the data used for training when it goes through the system? Does ZDR at Azure mean that in practice?
For all the major providers (except Deepseek), API use is simply never used for training (chat use often is).
My point (and the author's) was that even though they commit not to train, they do still retain data (security audits, etc).
With a ZDR agreement, they commit not to retain anything for any purpose.
If you look at opus on openrouter they have zdr on vertex and bedrock.
If you ask openrouter they will say that the vertex does not have the abuse monitoring opt out, ie it does retain data.
Bedrock there is no opt out, they just retain flagged content.
Both those endpoints on openrouter will human review flagged prompts.
The terms are in 'abuse monitoring'.