News for hackers. Stuff that matters.
Re:Gemini 3.5 Flash by simonw on Tuesday May 19, 2026 @03:29PM (#48198232)
The pelican is a lot: https://github.com/simonw/llm-gemini/issues/133#issuecomment...

Not a great bicycle though, it forgot the bar between the pedals and the back wheel and weirdly tangled the other bars.

Expensive too - that pelican cost 13 cents: https://www.llm-prices.com/#it=11&ot=14403&sel=gemini-3.5-fl...

Re:Gemini 3.5 Flash by hedgehog on Tuesday May 19, 2026 @03:32PM (#48198275)
That pelican looks like it's in Miami for a crypto conference.
Re:Gemini 3.5 Flash by joseda-hg on Tuesday May 19, 2026 @04:01PM (#48198672)
It looks like the starting soon screen of a crypto presentation
Re:Gemini 3.5 Flash by egillie on Tuesday May 19, 2026 @04:18PM (#48198925)
and somehow in 1992
Re:Gemini 3.5 Flash by xattt on Tuesday May 19, 2026 @03:53PM (#48198568)
It looks like it’s been partying for 60 years based on the wrinkles on its pouch.
Re:Gemini 3.5 Flash by ethbr1 on Wednesday May 20, 2026 @07:01AM (#48205821)
You don't know what that pelican has been through.
Re:Gemini 3.5 Flash by Xenoamorphous on Tuesday May 19, 2026 @04:25PM (#48199049)
Pelican in a white Testarossa.
Re:Gemini 3.5 Flash by verdverm on Tuesday May 19, 2026 @04:33PM (#48199171)
sorta looks like the Tron ripoff in the I/O keynote
Re:Gemini 3.5 Flash by brindleth on Tuesday May 19, 2026 @06:22PM (#48200472)
It look like the start of a new viral Peliwave aesthetic
Re:Gemini 3.5 Flash by airstrike on Tuesday May 19, 2026 @07:29PM (#48201074)
They're called ClawCons now
Re:Gemini 3.5 Flash by sho_hn on Tuesday May 19, 2026 @08:31PM (#48201551)
Personally, I don't attend them since I figured out I can set up agents to performatively engage in AI-related discussion and events for me, freeing up tons of my time thanks to automation.

Truly: Nothing better than AI tools to brave the challenges and requirements of modern life. "Claude, ride the hype train" is the decisive prompt you need.

Re:Gemini 3.5 Flash by coffeecoders on Tuesday May 19, 2026 @07:43PM (#48201182)
That pelican looks like it lost 100k on NFTs and now runs a paid stock-trading group.
Re:Gemini 3.5 Flash by seemaze on Tuesday May 19, 2026 @08:22PM (#48201469)
That pelican wears it's sunglasses at night. So it can, so it can keep track of the visions in it's eyes.
Re:Gemini 3.5 Flash by baochillchill on Tuesday May 19, 2026 @10:52PM (#48202549)
It looks quite funny.
Re:Gemini 3.5 Flash by whh on Tuesday May 19, 2026 @09:00PM (#48201742)
Pelican and I need an optometrist urgently
Re:Gemini 3.5 Flash by tantalor on Tuesday May 19, 2026 @04:01PM (#48198667)
Forgetting the chainstay is typical of asking random people to draw a bicycle.

https://www.gianlucagimini.it/portfolio-item/velocipedia/

> most ended up drawing something that was pretty far off from a regular men’s bicycle

Re:Gemini 3.5 Flash by et1337 on Tuesday May 19, 2026 @04:46PM (#48199381)
Asking random people to write SVG gives even worse results
Re:Gemini 3.5 Flash by lxgr on Tuesday May 19, 2026 @05:17PM (#48199784)
Especially without being able to look at the rendered output! (At least I'd be surprised if modern server-side tool calls regularly include an SVG renderer that can show a rasterized version to the model to iterate on it.)
Re:Gemini 3.5 Flash by gpm on Tuesday May 19, 2026 @09:19PM (#48201866)
One of the many things Google was pitching today is that they're going to run things like google search with access to linux container environments to do things like run tool calls... which will presumably be able to rasterize SVGs and show them to the model.

But Simon says he runs these through the API without tool access specifically to prevent that sort of "cheating". I.e. it's an LLM benchmark not an LLM+Harness benchmark.

Re:Gemini 3.5 Flash by Eji1700 on Tuesday May 19, 2026 @06:59PM (#48200800)
Although every single render of those has pedals on the correct side as opposed to the Gemini optical illusion back pedal that tries to be both on the other side of the central gear and infront of the back wheel.

Not really a criticism but an interesting point that you would never expect a human to make that mistake even in a bad drawing.

Re:Gemini 3.5 Flash by Barbing on Wednesday May 20, 2026 @01:17AM (#48203385)
Thanks for the delightful Velocipedia
Re:Gemini 3.5 Flash by smcleod on Tuesday May 19, 2026 @03:55PM (#48198597)
I feel like it embodies Google's vibe of an uncool guy trying to stay relevant to the youth pretty well.
Re:Gemini 3.5 Flash by dzhiurgis on Tuesday May 19, 2026 @09:42PM (#48202042)
That's grok. IMO both gemini and grok are the most overlooked models.
Re:Gemini 3.5 Flash by smcleod on Wednesday May 20, 2026 @04:24AM (#48204689)
Gemini is absolute garbage for anything useful, the last good model they released was 2.5 pro.
Re:Gemini 3.5 Flash by irthomasthomas on Tuesday May 19, 2026 @03:48PM (#48198494)
This is a perfect illustration of something I noticed with llm progress. Ask them to improve an svg like this, and it never fixes the missing crossbar or disconnected limbs, it just adds more stuff. In this example they have obviously improved greatly, and it contains a ridiculous amount of detail, but they still to get the basic shape of the frame right.
Re:Gemini 3.5 Flash by derefr on Tuesday May 19, 2026 @04:29PM (#48199105)
When you say "improve an svg like this", how are you imagining setting that workflow up? Are you just feeding them the SVG to iterate on; or are you giving them access to a browser to look at the rendering of the SVG?

I ask because:

Insofar as the original pelican test is zero-shot, it effectively serves as a way to test for the presence of a kind of "visual imagination" component within the layers of the model, that the model would internally "paint" an SVG [or PostScript, etc] encoding of an image onto, to then extract effective features from, analyze for fitness as a solution to a stated request, etc.

But if you're trying to do a multi-shot pelican, then just feeding back in the SVG produced in the previous attempt, really doesn't correspond to any interesting human capability. Humans can't take an SVG of a pelican and iteratively improve upon it just based on our imagined version of how that SVG renders, either! Rather, a human, given the pelican, would simply load the pelican SVG in a browser; look at the browser's rendering of the pelican; note the things wrong with that rendering; and then edit the SVG to hopefully fix those flaws (and repeat.)

I imagine current (mult-modal and/or computer-use) LLMs would actually be very good at such an "iterative rendered pelican" test.

Re:Gemini 3.5 Flash by irthomasthomas on Tuesday May 19, 2026 @04:39PM (#48199264)
I'm talking about two type of improvement, model improving, and prompt based improving. I am noticing that the baseline output has a lot more going on, the model has improved, yet it still makes those obvious looking mistakes with the shape of the frame or disconnected limbs etc.

And I am saying that if you take one of these SVGs and ask an LLM to look for flaws, it rarely spots those obvious flaws and instead suggests adding a sunset and fish in the birds mouth.

Re:Gemini 3.5 Flash by tskj on Wednesday May 20, 2026 @09:20AM (#48207268)
This is also my gripe with a lot of this stuff, always evaluating models on what they can literally oneshot is completely pointless; it's not how anything works, neither for humans nor for scaffolded AIs. I guess it's neat if you want to argue that a certain level of intelligence can "never be achieved" in a single forward pass, but like, so what. No one cares about that, except people who have already decided to be anti AI.

(not that I am in any sense pro AI, but it's just a weird lack of intellectual rigor)

Re:Gemini 3.5 Flash by irthomasthomas on Wednesday May 20, 2026 @10:40AM (#48208676)
Asking a model to improve its output is not one-shotting tho? My observation was that asking an llm to iterate and improve a response causes it to add more stuff, rather tha repair the broken stuff. And that model progress in general has the same pattern. This new model adds more details to its responses but continues to make mistakes at about the same rate.
Re:Gemini 3.5 Flash by losvedir on Wednesday May 20, 2026 @12:15PM (#48210099)
The question was whether you were giving it the rendered image and using the model's visual modal capability, or feeding back in the textual SVG.

It's hard to "imagine" what the rendered SVG looks like, for both humans and LLMs, so just iterating on text won't really be as useful of a test. But if you show it what it rendered, it might observe the bad-looking bicycle and be able to fix the text that way.

Re:Gemini 3.5 Flash by irthomasthomas on Wednesday May 20, 2026 @01:31PM (#48211159)
"I've even experimented with feeding the broken pelican svgs to an image model to look for flaws, and they still fail to spot the broken elements."
Re:Gemini 3.5 Flash by girvo on Tuesday May 19, 2026 @06:30PM (#48200571)
Their ability is best described as "spiky". To steal from aphyr: think kiki, more than bouba. Whats interesting is that a lot of the models seem to have similar spikes and "troughs", though there are differences.
Re:Gemini 3.5 Flash by stared on Tuesday May 19, 2026 @07:01PM (#48200825)
To a certain extent, it feels like a Sonnet 3.7 moment. Slightly overeager - you ask for a button color change, you see layout changes, new package dependencies, and the README rewritten from scratch - and not necessarily correctly.

When I ask for a pelican on a bike, I want the Platonic ideal of a pelican on a bike, not a vision of an alternative reality in which pelicans created bikes. Though, thinking about it again, maybe I should.

Re:Gemini 3.5 Flash by p1esk on Tuesday May 19, 2026 @08:30PM (#48201539)
What is “Sonnet 3.7 moment”?
Re:Gemini 3.5 Flash by stirfish on Tuesday May 19, 2026 @09:35PM (#48201986)
Slightly overeager - you ask for a button color change, you see layout changes, new package dependencies, and the README rewritten from scratch - and not necessarily correctly.
Re:Gemini 3.5 Flash by dormento on Wednesday May 20, 2026 @02:58PM (#48212394)
Sonnet 3.7 tried its damnedest but it was just kinda "off".
Re:Gemini 3.5 Flash by gowld on Tuesday May 19, 2026 @08:22PM (#48201478)
It's because LLMs are fundamentally generative (creative), not truth-seeking or logic-seeking. Simple logic has always been incredibly expensive to impossible for LLMs.
Re:Gemini 3.5 Flash by sosborn on Tuesday May 19, 2026 @09:00PM (#48201736)
This matches my experience with human too FWIW.
Re:Gemini 3.5 Flash by emp17344 on Tuesday May 19, 2026 @09:12PM (#48201828)
Why is there always an identical reply like this when anyone criticizes LLMs?
Re:Gemini 3.5 Flash by Araopa on Tuesday May 19, 2026 @10:05PM (#48202228)
So we have to train llms on debugging too, not just how to make things (which you easily train by feeding the outputs).
Re:Gemini 3.5 Flash by irthomasthomas on Monday June 01, 2026 @08:32AM (#48355980)
Same pattern for opus 4.7>4.8 and a 3D astronaut https://www.reddit.com/r/ClaudeAI/comments/1tt3a8h/differenc...
Re:Gemini 3.5 Flash by hydra-f on Tuesday May 19, 2026 @03:38PM (#48198347)
Same old issue with Gemini models trying to "enrich" everything
Re:Gemini 3.5 Flash by gcgbarbosa on Tuesday May 19, 2026 @03:53PM (#48198577)
funny that when I try the same prompt, gemini generates an image, not an SVG. something is not right.
Re:Gemini 3.5 Flash by simonw on Tuesday May 19, 2026 @04:09PM (#48198785)
That's likely because you're using the Gemini app which has a tool for image generation (nano banana) - I do my tests against the API to avoid any possibility of tool use.
Re:Gemini 3.5 Flash by nickmccann on Tuesday May 19, 2026 @04:19PM (#48198937)
This question makes me wonder if you one shot each pelican or do you run it a few times to get the best one?
Re:Gemini 3.5 Flash by simonw on Tuesday May 19, 2026 @06:43PM (#48200684)
I one-shot. I have a long-standing ambition to have each model generate 3x and then get the model (assuming it's a vision model) to pick the best one.
Re:Gemini 3.5 Flash by nashashmi on Tuesday May 19, 2026 @03:44PM (#48198431)
Beats a human by like 10$
Re:Gemini 3.5 Flash by unglaublich on Tuesday May 19, 2026 @03:56PM (#48198611)
So according to Google logic, the value of the pelican is $10-eps. (They applied that reasoning to conversions via adwords)
Re:Gemini 3.5 Flash by Barbing on Wednesday May 20, 2026 @01:21AM (#48203402)
Eps?
Re:Gemini 3.5 Flash by kirubakaran on Wednesday May 20, 2026 @01:52AM (#48203602)
epsilon
Re:Gemini 3.5 Flash by FranOntanaya on Wednesday May 20, 2026 @12:21PM (#48210169)
Only if you would use this pelican picture in production.
Re:Gemini 3.5 Flash by holtkam2 on Tuesday May 19, 2026 @03:55PM (#48198602)
at a certain point you're gonna need to change your benchmark because this will end up in the model's training set
Re:Gemini 3.5 Flash by simonw on Tuesday May 19, 2026 @04:09PM (#48198778)
Gemini were the team most likely to have this in their training set - see https://x.com/JeffDean/status/2024525132266688757 - and yet their latest model still messes up the bicycle frame!
Re:Gemini 3.5 Flash by recursive on Tuesday May 19, 2026 @05:58PM (#48200239)
I'm sure that certain point came and went many releases ago.
Re:Gemini 3.5 Flash by kzrdude on Wednesday May 20, 2026 @05:51AM (#48205321)
As mentioned in another recent thread, that time is now.
Re:Gemini 3.5 Flash by khy on Tuesday May 19, 2026 @04:56PM (#48199523)
That sun is very similar to the one from the background of this other top HN post about the OS museum: https://news.ycombinator.com/item?id=48195009
Re:Gemini 3.5 Flash by setgree on Tuesday May 19, 2026 @05:05PM (#48199633)
`<!-- Pelican Eye / Sunglasses (Cool Retro Aviators) -->`

wtf

`<!-- Gold Rim -->`

WTF??

Re:Gemini 3.5 Flash by nickvec on Tuesday May 19, 2026 @05:31PM (#48199958)
I enjoy the vaporwave aesthetic it went for. Looks like the pelican has a fish in its mouth too?

https://en.wikipedia.org/wiki/Vaporwave

Re:Gemini 3.5 Flash by __mharrison__ on Tuesday May 19, 2026 @06:03PM (#48200292)
They are just trolling you now
Re:Gemini 3.5 Flash by sbinnee on Tuesday May 19, 2026 @06:16PM (#48200420)
Wow what’s with all the styling? Is it manifestation of google’s styling bias? I like the result for sure. It’s shiny and pretty. But then it’s something I didn’t ask for.
Re:Gemini 3.5 Flash by TacticalCoder on Tuesday May 19, 2026 @06:30PM (#48200570)
Love your pelicans, as always. And that one is... Wow.

I noticed the "Synthwave" aesthetic, which is enjoying quite some success since quite some time now, has found its way into AI models (even when it's not in the user's query). It's not the first time I see the sun at sunset with color bands etc. in AI-generated pictures. Don't know why it's now taking on in AI too.

https://en.wikipedia.org/wiki/Synthwave

Hence the comments here about the 90s, Sonny Crockett's white Ferrari Testarossa in Miami, etc.

To be honest as a kid from the 80s and a teenager from the 90s who grew up with that aesthetic in posters, on VHS tape covers, magazine covers, etc. I do love that style and I love that it made a comeback and that that comeback somehow stayed.

Re:Gemini 3.5 Flash by kridsdale3 on Tuesday May 19, 2026 @07:42PM (#48201175)
Sythwave vibe hype hit a cultural high point with the release of Far Cry 3 Blood Dragon in 2013.

So it's as relevant and baked-in to today as actual 80s synth-culture was in 2000.

Re:Gemini 3.5 Flash by professoretc on Wednesday May 20, 2026 @12:31AM (#48203108)
"Look around to look around."
Re:Gemini 3.5 Flash by gowld on Tuesday May 19, 2026 @08:24PM (#48201490)
At the keynote today, Sundar Pichai asked Gemini to clone the Dino Game, and it added a synthwave-esque aesthetic.
Re:Gemini 3.5 Flash by danilocesar on Tuesday May 19, 2026 @08:18PM (#48201424)
Given your pelican is very famous now, don't you think they are adding instructions to beat this benchmark those days?
Re:Gemini 3.5 Flash by Culonavirus on Tuesday May 19, 2026 @08:22PM (#48201476)
Well clearly it's not working lmao
Re:Gemini 3.5 Flash by tandr on Tuesday May 19, 2026 @08:18PM (#48201434)
If you sort that table by "output token price", it gets really terrifying - going from 4 cents up to $600 =8-O
Re:Gemini 3.5 Flash by nrds on Tuesday May 19, 2026 @08:27PM (#48201515)
We've been daily-driving this model for a few weeks and let me tell you, everything it does is a lot. Fast as fuck and it's actually not bad intelligence-wise for a fast model. It basically tries to make up for any intelligence deficit by just doing a lot, checking a lot, retrying a lot.

That's not to say I don't spend my days raging at it... a lot... but it's not that bad. It does tend to ignore completion criteria but it doesn't obviously degrade when being nudged like some models do.

Re:Gemini 3.5 Flash by bitexploder on Wednesday May 20, 2026 @04:30AM (#48204721)
One time I told it “we are doing science” and I had DNA emoji everywhere and it so over enthusiastically embraced the science theme I was genuinely laughing. It finished one task with a flourish of several dna emoji and proclaimed: The Science is COMPLETE. I died.

It really is a lot some of the time. And it’s chain of thought is hilarious a lot of the time.

Re:Gemini 3.5 Flash by taurath on Tuesday May 19, 2026 @08:40PM (#48201603)
I can’t help but think that what AI is best at is convincing management that things it creates are full featured which reads to their brains as mature
Re:Gemini 3.5 Flash by dekhn on Tuesday May 19, 2026 @08:55PM (#48201701)
I'm told there is a new Jeff Dean fact inside google: "Jeff Dean manually adjusts the weights in the model just to screw with Simon".
Re:Gemini 3.5 Flash by karmakaze on Tuesday May 19, 2026 @09:01PM (#48201748)
I'm hoping we'll have many of these pelican cyclist pictures collected. Then when all the models can do it well, we'll stop posting about them, and dhen the next generations of AIs train on the data we'll have these canonical archetypes.
Re:Gemini 3.5 Flash by Razengan on Tuesday May 19, 2026 @09:56PM (#48202151)
I've found prompts like "capybara with spotted fur and 7 octopus tentacles instead of legs, each a different color, riding a tricycle" etc. to be a better test

Last time I tried, ChatGPT's image generator got the best result.

Re:Gemini 3.5 Flash by bee_rider on Tuesday May 19, 2026 @10:22PM (#48202338)
I wonder if they added all these unrequested details as an Easter-egg or something? (Since they must be aware of your test by now).
Re:Gemini 3.5 Flash by VectorLock on Wednesday May 20, 2026 @12:40AM (#48203174)
The fact it went for vaporwave styling on its own is very telling.
Re:Gemini 3.5 Flash by dankwizard on Wednesday May 20, 2026 @01:06AM (#48203328)
Wouldn't be a thread about the tech that is changing the landscape for businesses across nearly every discipline without a pelican svg.
Re:Gemini 3.5 Flash by nomilk on Wednesday May 20, 2026 @05:43AM (#48205262)
'Pelicans' should be the unit of measurement for model prices, rather than tokens.