Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: AI Playground by Vercel Labs

play.vercel.ai

Post reply on HN
Print view View original

1–10 of 103 posts

Show HN: AI Playground by Vercel Labs

#1

Post by palmdeezy »

Hey, Jared Palmer (creator of this playground) here. Really excited to ship this. I’ve been building this over the past few weeks to compare LLMs from different providers like OpenAI, Anthropic, Cohere, etc. At Vercel, I manage our Frameworks division (including Next.js, Svelte, and Turbo) and wanted to also dogfood some of the latest features in a slightly larger application. This playground takes a lot of inspiration from https://nat.dev and is built on Tailwind, ui.shadcn.com, and some upcoming Vercel products we’re announcing soon. We’re going to continue adding models to compare and add other frameworks to generate code snippets from.

Show HN: AI Playground by Vercel Labs
play.vercel.ai

Re: Show HN: AI Playground by Vercel Labs

#2

Post by doodlesdev »

Please provide an icon that makes sense on other platforms for the shortcut on the right bottom, not everyone uses Macs! Overall, pretty cool idea, how is Vercel gonna handle API costs for this? I imagine this is pretty expensive to run, specially considering GPT4 access.

Also, I have to link this here because I thought it was too funny:

https://play.vercel.ai/r/G9y3eXJ

edit: Oh I just hit the token limit, I guess that's how you're dealing with cost lol. Something I was gonna suggest is that maybe answers should be cached to prevent repeated calls to the same API when the prompt hasn't changed, this could potentially increase the amount of tests I could do.

Re: Show HN: AI Playground by Vercel Labs

#3

Post by steventey »

doodlesdev wrote: post #2

Please provide an icon that makes sense on other platforms for the shortcut on the right bottom, not everyone uses Macs! Overall, pretty cool idea, how is Vercel gonna handle API costs for this? I imagine this is pretty expensive to run, specially considering GPT4 access. Also, I have to link this here because I thought it was too funny: https://play.vercel.ai/r/G9y3eXJ edit: Oh I just hit the token limit, I guess th…

"How many words are there in "The Star Spangled Banner"?" LOL hahaha

Will relay your feedback about the icon back to the team - thank you for trying this out!

Re: Show HN: AI Playground by Vercel Labs

#7

Post by swyx »

beautiful work as always jared!

one of the useful things from nat.dev was the tokens/second calculation, latency is impt as usual. would be nice to enable?

i'm excited to see what the upcoming products are. making things default sharable like this across all frameworks would be really nice.

Re: Show HN: AI Playground by Vercel Labs

#8

Post by palmdeezy »

swyx wrote: post #7

beautiful work as always jared! one of the useful things from nat.dev was the tokens/second calculation, latency is impt as usual. would be nice to enable? i'm excited to see what the upcoming products are. making things default sharable like this across all frameworks would be really nice.

Thanks man! Good idea on token/s. Will add

Re: Show HN: AI Playground by Vercel Labs

#9

Post by tough »

palmdeezy wrote: post #8
swyx wrote: post #7

beautiful work as always jared! one of the useful things from nat.dev was the tokens/second calculation, latency is impt as usual. would be nice to enable? i'm excited to see what the upcoming products are. making things default sharable like this across all frameworks would be really nice.

Thanks man! Good idea on token/s. Will add

Would be nice to know cost too!

tokmon is on showHN today too

https://news.ycombinator.com/item?id=35616871

Re: Show HN: AI Playground by Vercel Labs

#10

Post by cramforce »

palmdeezy wrote: post #8
swyx wrote: post #7

beautiful work as always jared! one of the useful things from nat.dev was the tokens/second calculation, latency is impt as usual. would be nice to enable? i'm excited to see what the upcoming products are. making things default sharable like this across all frameworks would be really nice.

Thanks man! Good idea on token/s. Will add

I'm super excited in graphing this over time. Will be interesting to see how the providers develop over the coming months. FWIW we also need TTFB not just tokens per second. The first token is key for UX
Post reply on HN