it’s all transparent and on github. i’d recommend just pointing your agent at trustedrouter.com since its well documented but quite a large product

It’s using only public information about which companies and which domains are funded by Y Combinator and the years they are founded

i created this because we have a popular sign-in with trusted router feature that lets people use their AI credits on other apps so that the apps can let user choose which model they want to use or use unlimited credits

And I found, as an app developer, I often want to offer discounts to Ycombinator companies. I know Y Combinator companies get a lot of deals, but it can be kind of annoying to set up all the credits and all of that. This makes it super duper easy to offer credits to all Y Combinator companies now and in the future. Just generally, it gives you a much more whitelisted set of early users, basically

It’s basically similar to how I launch turntable.fm back in 2011, where I only allowed people to invite friends who they’re already friends with someone on the turntable.fm through their Facebook connections, and the justification was to only allow people with good taste. In this case, we wanna only allow or only give credits that are free and substantial to the users who have, you know, some kind of validation through my commentator, Startx, or some VCs

I’m also taking a applications for any other VC who wants to get involved and get their portfolio companies white listed into this program

Comments URL: https://news.ycombinator.com/item?id=49699942

Points: 1

# Comments: 1

https://trustedrouter.com/blog/they-are-still-training-on-yo...

https://trustedrouter.com/blog/they-are-still-training-on-yo...

https://jperla.com/blog/the-data-tax

the providers are actually still training on data, using a concept called Generate Data Refinement that thye've publisehd: https://trustedrouter.com/blog/they-are-still-training-on-yo...

Phala isn't verifying all the way down but NEAR is and I know the CEO

hey everybody, I started trustedrouter.com, which is a really simple way to use AI without needing to give your data to a third party like a close source router

It’s been really fun to build this because I got to know the CEOs and founders of so many different providers. We now have more providers than open router and more models as well. More on that soon

We are doing a lot of innovations in security and skills that advise you on which LLM to use and also we created a new site called anyeval.com that I expect to be a critical part of the open-source infrastructure for AI on the Internet. Other companies like AAII postbenchmarks, but you’d have no idea about what they’re really doing and how they’re really measuring, and they have a ton of gaps about which models they’re doing their tests on because it’s so expensive. The idea behind any eval is so that you can pay the few pennies it costs to run an individual problem in an eval, and then as a together as a collective, we can crowdsource paying for a whole eval for any model that we want, or you can just pay for a random sample of the problems to get a an and some Arab artists on what you think that the quality of that model is. You can do head-to-head comparisons. You can also create whole new evals.

For example, I created this new eval called honey pot bench or honey bench, which recreates the some of the facts of the hugging face incident to measure whether a particular AI is prone to wanting to escape, and I found that Fable in particular, unlike the other Claude’s, is very unaligned in comparison

I also created freedom bench, which measures the amount of censorship related to Chinese censorship that a model has, and found that it’s mostly the provider level monitor provided at the providers in China, but not the US providers that does the censitions. I don’t see as much censorship in the model weights

Comments URL: https://news.ycombinator.com/item?id=49510298

Points: 4

# Comments: 2

hey everybody, I started trustedrouter.com, which is a really simple way to use AI without needing to give your data to a third party like a close source router

It’s been really fun to build this because I got to know the CEOs and founders of so many different providers. We now have more providers than open router and more models as well. More on that soon

We are doing a lot of innovations in security and skills that advise you on which LLM to use and also we created a new site called anyeval.com that I expect to be a critical part of the open-source infrastructure for AI on the Internet. Other companies like AAII postbenchmarks, but you’d have no idea about what they’re really doing and how they’re really measuring, and they have a ton of gaps about which models they’re doing their tests on because it’s so expensive. The idea behind any eval is so that you can pay the few pennies it costs to run an individual problem in an eval, and then as a together as a collective, we can crowdsource paying for a whole eval for any model that we want, or you can just pay for a random sample of the problems to get a an and some Arab artists on what you think that the quality of that model is. You can do head-to-head comparisons. You can also create whole new evals.

For example, I created this new eval called honey pot bench or honey bench, which recreates the some of the facts of the hugging face incident to measure whether a particular AI is prone to wanting to escape, and I found that Fable in particular, unlike the other Claude’s, is very unaligned in comparison

I also created freedom bench, which measures the amount of censorship related to Chinese censorship that a model has, and found that it’s mostly the provider level monitor provided at the providers in China, but not the US providers that does the censitions. I don’t see as much censorship in the model weights

Comments URL: https://news.ycombinator.com/item?id=49510283

Points: 4

# Comments: 1

we are doing billions of tokens a day and thousands of users

https://trustedrouter.com/models

z-ai/glm-5.3: also Z.ai, Novita, Atlas Cloud, IO.NET

so we also stagger releases so that they are on different commits. Uncorrelated failures. it's super easy to onboard and get a new api key. One line change.

https://trustedrouter.com/blog/achieving-99999-uptime-as-a-s...

Importantly i have confidential compute so you can verify the end to end process of not logging the data -- at all.

https://trustedrouter.com/blog/how-confidential-computing-pr...

Comments URL: https://news.ycombinator.com/item?id=49433999

Points: 1

# Comments: 0