Frankie Chat has all the latest models from OpenAI, like GPT 4.1, o3 and o4-mini.

We also have DeepSeek V3 and R1, Grok 3 and starting today Anthropic Claude 4 sonnet and 3.5 haiku.

Try it out: signin with nostr, recharge with sats: https://www.frankie.tools

Reply to this note

Please Login to reply.

Discussion

4 https://www.frankie.tools

haiku.

Try out: models like 3 4.1, sats: from V3 recharge o4-mini. it starting today Anthropic latest and OpenAI, all R1, 3.5 have GPT with and Claude o3 Grok the with

We signin also nostr, has Frankie and DeepSeek Chat and sonnet

Did you offer an oai like api?

No :(

You think I should? What is the use case?

Plug your service in any app, for example in my IDE or any other app

Makes sense, but why not connect the actual APIs?

They all have basically the very same HTTP interface. The burden is just to keep multiple API keys...

But yeah I see that it could be convenient to just pay with sats with no accounts, keys or else...

Exactly. It's the openrouter value proposal. One api, all models, much better if you can remain anonymous and pay with sats

Yes, I see it.

I have to investigate potential "legal" problems I can bump into.

In the chat all user messages are filtered by the moderation service, so I am cool with that.

I will think about it. Thanks for sharing!

Thanks you!

Yes, if you do that more people will be able to use your service

How do you implement Privacy given that the payment is linked to a Nostr npub??

The quick answer: I do not, and I do not think I have to.

People can choose to just have an npub dedicated to services like Frankie, different from the one used here on the "social" nostr.

It is quite simple with Amber or Alby to manage multiple identities and that's the best practice anyway.

Not sure how I could do it without making the UX much worst.

Very Clear Thks

One last question. How about the pricing?

They are priced against the actual model pricing exposed by OpenAI and the rest.

With some surplus, obviously, that I keep.

The best info I can give is my own "bill" as a user since August 5 2023 when it got live. Purchased 40 USD, spent 26.

That's 26 USD in almost 22 months of usage, and I use it pretty much, for coding and what not.

Your mileage may vary but we're talking just over 1USD per month :)

Looks cool, if the actual use is frequent enough. Average number of prompt? Approximately..?

I do not have stats on hand, my bad.

But I can go from 2-3 conversation per day on a lousy day, to 12-14 on an average productive one.

Each conversation has multiple prompts.

I am adding the web search models today, I will try and get more stats on usage "soon" :)

Great

Oh, important note: image generation is *much* more expensive.

It starts from around 40 cents for the lower resolution, using the last OpenAI model.

I almost never do images, so my rates are based on text chat exclusively.