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

Getting a Gemini API key is an exercise in frustration

ankursethi.com

Post reply on HN
Print view View original

171–180 of 361 posts

Re: Getting a Gemini API key is an exercise in frustration

#171

Post by rand17 »

So you don't like writing "the boring code". What do you expect from writing a CRUD? What would you like to write? What "interesting problems" would you like to focus on? Great sadness will fall upon the industry when the last graybeard dies, who had this arcane knowledge of "writing code". I have bad Player Piano vibes nowdays.

Around me devs are beginning to warm up to the idea, that they are not coders (and neither should I be), but "prompt engineers". When I take too much time on a task, when I can't solve a problem with a push of a button, when I muse about copilot hallucinations in my PR - someone usually comes helpfully to tell me, I need better prompting skills. Have you tried this expression? Have you tried more context? Have you tried with this copy pasted magical formula?

No creative worker in human history was so overjoyed to devalue his or her work and knowledge in such haste.

Re: Getting a Gemini API key is an exercise in frustration

#172

Post by ianberdin »

CSMastermind wrote: post #147
Ozzie_osman wrote: post #145

I was recently (vibe)-coding some games with my kid, and we wanted some basic text-to-speech functionality. We tested Google's Gemini models in-browser, and they worked great, so we figured we'd add them to the app. Some fun learnings: 1. You can access those models via three APIs: the Gemini API (which it turns out is only for prototyping and returned errors 30% of the time), the Vertex API (much more stable but lac…

Some other fun things you'll find: - The models perform differently when called via the API vs in the Gemini UI. - The Gemini API will randomly fail about 1% of the time, retry logic is basically mandatory. - API performance is heavily influenced by the whims of the Google we've observed spreads between 30 seconds and 4 minutes for the same query depending on how Google is feeling that day.

Even funnier, when Pro 3 answers to a previous message in my chat. Just making a duplicate answer with different words. Retry helps, but…

Re: Getting a Gemini API key is an exercise in frustration

#173

Post by kkarpkkarp »

Aeolun wrote: post #168

This is exactly my experience with gemini, and exactly why I bounced on the stupid thing. I just don’t have hours to waste on Google’s stupid processes.

And exactly same here. I wonder how much money they lose because of this poor process.

I am not going to use Gemini API in foreseen future as I don't want to manage those keys anymore. No matter how good their model is

Re: Getting a Gemini API key is an exercise in frustration

#174

Post by mittensc »

rand17 wrote: post #171

So you don't like writing "the boring code". What do you expect from writing a CRUD? What would you like to write? What "interesting problems" would you like to focus on? Great sadness will fall upon the industry when the last graybeard dies, who had this arcane knowledge of "writing code". I have bad Player Piano vibes nowdays. Around me devs are beginning to warm up to the idea, that they are not coders (and neithe…

Other side of the equation:

I remember learning C++ with something like valgrind. I would write stupid code, validate, fix stupid issues.

Others before me learned the harder way.

With LLMs right now I'm learning frontend by just generating the UIs I want.

I'm getting the code/mocks and experimenting.

It's bad code, i will need to adjust, but it helps immensely as a starting point same as valgrind helped in the past.

Trying to learn via searching for info just doesn't work as well with all the flood of spam.

Re: Getting a Gemini API key is an exercise in frustration

#175

Post by ozozozd »

jiggawatts wrote: post #128
sofixa wrote: post #37

Earlier quoted context omitted.

Baring them using specific marketing terms (so you have EC2 for what are basically virtual machines), for which both the docs and the portal itself provide helpful information, what do you mean? I find GCP's console and whole set up to be slightly better, but both it and AWS are fine. Now Azure, or anything made by VMware, you just know they hate you.

I love how people think Azure is somehow worse than AWS when the latter isn't even a single portal, it's many , each of which shows just one product in one region. Oh, you needed a VM with a network and some storage, including access to blobs somewhere else in the world? Just open up a dozen tabs and join the randomly generated gibberish resource identifiers yourself manually like a savage!

Exactly. Just as the lord intended.

Although Azure just randomly fails, and then it turns out it actually worked but the UI had failed. But then the next step throws an obscure error message, but you get around that on a different screen, so on so forth…

Re: Getting a Gemini API key is an exercise in frustration

#177

Post by rozap »

I wasted several hours this week going around in the exact same circles. We have a billing account, but kept hitting a gemini quota. Fine. But then on the quota page, every quota said 0% usage. And our bill was like $5. Some docs said check AI studio, but then the "import project from google cloud to AI studio" button kept silently failing. This was a requests per minute quota, which was set at 15 (not a whole lot...) but wouldn't reset for 24 hours. So then I kept making new projects so I could keep testing this thing I'm building, until eventually I ran out.

The only way we could get it resolved was to (somehow) get a real human at google on the phone because we're in some startup program or something and have some connection there. Then he put in a manual request to bump our quota up.

Google cloud is the most kafkaesque insane system I've ever had the misfortune of dealing with. Every time I use it I can tell the org chart is leaking.

Re: Getting a Gemini API key is an exercise in frustration

#178

Post by specproc »

CSMastermind wrote: post #147
Ozzie_osman wrote: post #145

I was recently (vibe)-coding some games with my kid, and we wanted some basic text-to-speech functionality. We tested Google's Gemini models in-browser, and they worked great, so we figured we'd add them to the app. Some fun learnings: 1. You can access those models via three APIs: the Gemini API (which it turns out is only for prototyping and returned errors 30% of the time), the Vertex API (much more stable but lac…

Some other fun things you'll find: - The models perform differently when called via the API vs in the Gemini UI. - The Gemini API will randomly fail about 1% of the time, retry logic is basically mandatory. - API performance is heavily influenced by the whims of the Google we've observed spreads between 30 seconds and 4 minutes for the same query depending on how Google is feeling that day.

I have also had some super weird stuff in my output (2.5-flash).

I'm passing docs for bulk inference via Vertex, and a small number of returned results will include gibberish in Japanese.

Re: Getting a Gemini API key is an exercise in frustration

#179

Post by rand17 »

mittensc wrote: post #174
rand17 wrote: post #171

So you don't like writing "the boring code". What do you expect from writing a CRUD? What would you like to write? What "interesting problems" would you like to focus on? Great sadness will fall upon the industry when the last graybeard dies, who had this arcane knowledge of "writing code". I have bad Player Piano vibes nowdays. Around me devs are beginning to warm up to the idea, that they are not coders (and neithe…

Other side of the equation: I remember learning C++ with something like valgrind. I would write stupid code, validate, fix stupid issues. Others before me learned the harder way. With LLMs right now I'm learning frontend by just generating the UIs I want. I'm getting the code/mocks and experimenting. It's bad code, i will need to adjust, but it helps immensely as a starting point same as valgrind helped in the past.…

I do not think that all LLMs are evil; they are valid tools, but it's not a hammer with meta glasses attached to render everything into a nail. I also find it very useful in certain situations - but not in all situations.

Two more things. Bad code (in work, in reality, not in a hobby project) is rarely converted into good code. And the last one: in my twenty plus years of being a dev, this is the first year job offers simply just dried up. With bad code being good enough (hey, it compiles! it mostly works!), hopefully you and I will be the lucky few to still be in the business five years later.

Re: Getting a Gemini API key is an exercise in frustration

#180

Post by bobjordan »

I went through similar song and dance using a paid Gemini code assist “standard” level subscription. I finally got Gemini 3 working in my terminal in my repository. I assigned it a task that Claude code Opus 4.5 would quickly knock out, and Gemini 3 did a reasonably similar job. I had opus 4.5 evaluate the work and it was complimentary of Gemini 3S work. Then I check the usage and I’d used 10% of the daily token usage limit, about 1.5M tokens on that one task. So I can only get about 10 tasks before I’m rate limited. Meanwhile with Claude code $200 max plan, I can run 10 of those same caliber of tasks in parallel, even with opus 4.5 model, and barely register the usage meter. The only thing the Gemini code assist “standard” plan will be good for with these limits are just double checking the plans that opus 4.5 makes. Until the usage limits are increased, it’s pretty useless compared to Claude code max plan. But there doesn’t seem to be any similar plan offering from Google.
Post reply on HN