I used this just-released API (of Gemini Pro) with multimodal input to test some of the things from the infamous Gemini Demo. You can see here [ <a href="https://www.youtube.com/watch?v=__nL7Vc0OCg" rel="nofollow noreferrer">https://www.youtube.com/watch?v=__nL7Vc0OCg</a> ] my GPT-4 recreation of that ad which went viral.<p>Gemini Pro is... not great. In one test, I asked what gesture I was making (while showing a thumbs up) -- it said thumbs down and "The image is a commentary on the changing nature of truth".<p>I just just made a heads-to-heads comparison -- you can watch it here: <a href="https://www.youtube.com/watch?v=1RrkRA7wuoE" rel="nofollow noreferrer">https://www.youtube.com/watch?v=1RrkRA7wuoE</a><p>Code is here: <a href="https://github.com/gregsadetsky/sagittarius">https://github.com/gregsadetsky/sagittarius</a>
Free quota looks reasonable with 60 queries per minute. On the other hand data from free quota requests will be used to improve the product.<p><a href="https://ai.google.dev/pricing" rel="nofollow noreferrer">https://ai.google.dev/pricing</a>
I know it's just an anecdote, but my biggest problem with Google's Bard/Gemini is that the moment I tried to ask a question about something - I started getting ads all over the internet and social media related to that.<p>Doing this with ChatGPT 4.0 for months and months did not cause this type of behavior.
I like that they have a "blog post creator"[1] in their examples. There's no hope for the future of the web when the self-proclaimed stewards of its quality encourage AI spam.<p>[1] <a href="https://makersuite.google.com/app/prompts/blog-post-creator" rel="nofollow noreferrer">https://makersuite.google.com/app/prompts/blog-post-creator</a>
Developers can start building with our first version of Gemini Pro through Google AI Studio at ai.google.dev<p>Developers have a free quota and access to a full range of features including function calling, embeddings, semantic retrieval, custom knowledge grounding, chat functionality and more. It supports 38 languages across 180+ countries.
> Access restricted
You do not have permission to view this page.<p>Wait only in the US?<p>Edit: I can access it through the Google Cloud Console.<p><a href="https://imgur.com/a/NXAgvFb" rel="nofollow noreferrer">https://imgur.com/a/NXAgvFb</a>
Some thoughts comparing this to the GPT API (from a thread: <a href="https://hachyderm.io/@ianbicking/111574983914336748" rel="nofollow noreferrer">https://hachyderm.io/@ianbicking/111574983914336748</a>):<p>It looks like a fairly easy swap-in for GPT. "messages" becomes "content". Some of the configuration parameters are slightly different (topP/etc), but I have never put in the effort to understand the practical effect of those so I never tweak their values.<p>The messages themselves are a list of "parts", which allows mixed media messages. This feels a little cleaner than how GPT has handled messages being extended.<p>Instead of role: "assistant" they use role: "model". There's no role: "system" – presumably you just shove everything into user messages. You can also leave off the role... and I assume that means default to "user" but it's not clear if it's 100% equivalent...?<p>There's a bunch of moderation parameters, which seems like a good idea. OpenAI has a moderation endpoint you can use to preflight check your input, but doing it all at once makes more sense. There's four categories and you can adjust your sensitivity to each (and turn off blocking at entirely). The sensitivity is not about how extreme the violation is, but how likely it is a violation. So it's not like a G/PG/PG-13/etc rating. Just a question of how many false positives/negatives you want.<p>There's functions, though they are in beta (whatever that means): <a href="https://ai.google.dev/docs/function_calling" rel="nofollow noreferrer">https://ai.google.dev/docs/function_calling</a> – they look very very similar to GPT functions. They don't have the "JSON response" that GPT has, but that seems mostly redundant with functions anyway.<p>I have no idea how well prompts translate, but it feels like the API is an easy translation. And importantly everything is semantically equivalent, you don't have to make one pretend it is the other, like turning a completion API into a chat API.<p>Given the generous free tier I feel fairly motivated to swap in Gemini and try to ship experiments that I've sat on until now.
why on earth did they design the Node.js and Web APIs to be slightly different and incompatible? (edit: this might just be a bug/oversight on the landing page?)<p>Node.js:<p><pre><code> const model = genAI.getGenerativeModel({ model: "gemini-pro-vision"});
const result = model.generateContent({ contents: [{parts: [
{text: "What’s in this photo?"},
{inlineData: {data: imgBase64, mimeType: 'image/png'}}
] }] })
</code></pre>
Web:<p><pre><code> const model = genAI.getGenerativeModel({ model: "gemini-pro-vision"});
const result = await model.generateContent([
"What’s in this photo?",
{inlineData: {data: imgDataInBase64, mimeType: 'image/png'}}
]);</code></pre>
You can make 1 query per second to it for free, including large queries that contain images ? This is crazy !<p>I will happily let google buy me for that price.<p><a href="https://ai.google.dev/pricing" rel="nofollow noreferrer">https://ai.google.dev/pricing</a>
Wow. what a crap site.
I clicked on the option for prompt thinking I could go back and request an API key. Boy was I wrong. No matter what I do it takes me to the prompt console where I get Access Denied and it hijacked my back button.
When I try to create an API key it says that "We are sorry, but you do not have access to Early Access Apps" yet my domain does allow access to early access apps....
See how to use new Gemini Embeddings with Qdrant Vector Database
<a href="https://qdrant.tech/documentation/embeddings/gemini/" rel="nofollow noreferrer">https://qdrant.tech/documentation/embeddings/gemini/</a>
I'd like to know why the name of this AI product coincides with the alternative in-between-HTTP-and-Gopher Gemini protocol.<p>I'm sure it's just an accident.
I can only access <a href="https://makersuite.google.com/" rel="nofollow noreferrer">https://makersuite.google.com/</a> when using a VPN to the US. Also, it spams popups that get blocked by Firefox.<p>Some basic prompts, which are answered correctly most of the time by ChatGPT4:<p>There are 31 books in my house. I read 2 books over the weekend. How many books are still in my house?<p>> 29 books<p>Julia has three brothers, each of them has two sisters. How many sisters does Julia have?<p>> Three<p>If you place an orange below a plate in the living room, and then move the plate to the kitchen, where is the orange now?<p>> Under the plate in the kitchen.<p>So, not great.
Typical Google UX.<p>Get API key, takes me to makersuite, where I get a create API key button that errors. Then when I reload the page, I get a straight forbidden page.<p>HP said it best, you have to isolate the team from the bigger company to allow them to work as an effective startup. How can solo-preneurs provide better UX & onboarding while doing 16 other jobs than Google can with multi-billion dollar budgets?
Cross posting some links from another post that HNers found helpful<p>- <a href="https://cloud.google.com/vertex-ai" rel="nofollow noreferrer">https://cloud.google.com/vertex-ai</a> (marketing page)<p>- <a href="https://cloud.google.com/vertex-ai/docs" rel="nofollow noreferrer">https://cloud.google.com/vertex-ai/docs</a> (docs entry point)<p>- <a href="https://console.cloud.google.com/vertex-ai" rel="nofollow noreferrer">https://console.cloud.google.com/vertex-ai</a> (cloud console)<p>- <a href="https://console.cloud.google.com/vertex-ai/model-garden" rel="nofollow noreferrer">https://console.cloud.google.com/vertex-ai/model-garden</a> (all the models)<p>- <a href="https://console.cloud.google.com/vertex-ai/generative" rel="nofollow noreferrer">https://console.cloud.google.com/vertex-ai/generative</a> (studio / playground)<p>VertexAI is the umbrella for all of the Google models available through their cloud platform.<p>You want the last link if you are looking for a ChatGPT like experience, with the ability to also adjust the parameters, so more like a UI on top of the API
When I enter into the AI, Firefox blocks an insane amount of popups. The counter for blocked pop ups quickly reaches >100 where it stops counting. What is it trying to do?
This is very good:<p>- 60 queries per minute free
- about 1/5th the price of GPT3.5 Turbo
- priced per char, not per token
- same image pricing as GPT4 150x150
We put the image portion through its paces and compared it with GPT-V here: <a href="https://blog.roboflow.com/first-impressions-with-google-gemini/">https://blog.roboflow.com/first-impressions-with-google-gemi...</a>