frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

The chemical attack that forced the Northern line's extraordinary repair

https://www.ianvisits.co.uk/articles/the-hidden-chemical-attack-that-forced-the-northern-lines-ex...
1•zeristor•1m ago•0 comments

Meta Execs Privately Compared Instagram to Addictive Drug, Court Filing Shows

https://www.nationalreview.com/news/meta-researchers-privately-compared-instagram-to-addictive-dr...
2•fortran77•2m ago•1 comments

After nearly 100 years, scientists may have detected dark matter

https://www.u-tokyo.ac.jp/focus/en/press/z0508_00433.html
1•geox•2m ago•0 comments

AI Agent that automates biotech experiments

https://twitter.com/liuwilliam47/status/1993474101152039032
1•WilliXL•4m ago•0 comments

Inside the DOGE Succession Drama Elon Musk Left Behind

https://www.politico.com/news/magazine/2025/11/21/doge-elon-musk-succession-00641110
2•tastyface•5m ago•1 comments

CS234: Reinforcement Learning Winter 2025

https://web.stanford.edu/class/cs234/
1•jonbaer•8m ago•0 comments

Doge May Be Dead, but Its Crimes Live On

https://newrepublic.com/article/203638/doge-dead-rest-in-piss
5•doener•8m ago•0 comments

NJ company is allowed to import cocaine plants (2023)

https://nj1015.com/this-nj-company-is-allowed-to-import-cocaine-plants/
1•raw_anon_1111•9m ago•0 comments

Brand New Layouts with CSS Subgrid

https://www.joshwcomeau.com/css/subgrid/
1•soheilpro•10m ago•1 comments

Show HN: Tripnly Lisboa City Pass – 5 Days for 1€

https://lisboacitypass.tripnly.com
2•alper_aydin•10m ago•0 comments

What Now? Handling Errors in Large Systems

https://brooker.co.za/blog/2025/11/20/what-now
1•thundergolfer•10m ago•0 comments

An Incomplete List of Eminent Psychologists of the Modern Era (2014)

https://psycnet.apa.org/fulltext/2014-35304-001.html
1•Sir_Twist•11m ago•0 comments

I Got Working RCS Messaging

https://wt.gd/working-rcs-messaging
4•joecool1029•17m ago•0 comments

Secrets in unlisted GitHub gists are now reported to secret scanning partners

https://github.blog/changelog/2025-11-25-secrets-in-unlisted-github-gists-are-now-reported-to-sec...
1•PieUser•18m ago•0 comments

Mass Surveillance Is Powering a New Era of Pretextual Traffic Stops

https://reason.com/2025/11/24/mass-surveillance-is-powering-a-new-era-of-pretextual-traffic-stops/
3•pseudolus•20m ago•0 comments

Ask HN: How does one move from BigTech to more fullfilling places?

3•conqrr•22m ago•1 comments

Looking to build muscle? Lose weight? Need more protein, right? Probably not

https://news.harvard.edu/gazette/story/2025/11/looking-to-build-muscle-lose-weight-need-more-prot...
2•gnabgib•26m ago•0 comments

Tell HN: Karpathy's MinGPT Works on Windows and AMD (Ryzen Max)

1•lostmsu•26m ago•0 comments

Show HN: Parm – Install GitHub releases just like your favorite package manager

https://github.com/yhoundz/parm
1•houndz•27m ago•0 comments

The gruesome new data on tech jobs

https://www.businessinsider.com/gruesome-tech-jobs-data-scientists-analytics-indeed-2025-11
9•pseudolus•27m ago•1 comments

Brain changes observed during pregnancy (2024)

https://www.nih.gov/news-events/nih-research-matters/brain-changes-observed-during-pregnancy
1•thunderbong•31m ago•0 comments

Learn to code music in the browser with strudel.cc (Free and open-source)

https://terryds.notion.site/Learning-Music-with-Strudel-2ac98431b24180deb890cc7de667ea92
1•terryds•32m ago•0 comments

Dear Peter (Thiel)

https://andys.blog/dear-peter/
1•andytratt•39m ago•2 comments

Daily coffee drinking may slow bio aging of people with major mental illness

https://medicalxpress.com/news/2025-11-daily-coffee-biological-aging-people.html
4•bikenaga•39m ago•1 comments

Unix man pages: AI-friendly documentation since 1971

https://www.nutrient.io/blog/unix-man-pages-ai-friendly-documentation/
2•ingvarr•40m ago•0 comments

Meta had a 17-strike policy for sex trafficking, former safety leader claims

https://www.theverge.com/news/827658/meta-17-strike-policy-sex-trafficking-testimony-lawsuit
5•randycupertino•43m ago•0 comments

SF home invasion targets crypto of Physical Intelligence co-founder Lachy Groom

https://nypost.com/2025/11/25/us-news/sam-altmans-ex-boyfriends-san-francisco-home-targeted-in-cr...
4•pants2•44m ago•2 comments

Show HN: I Figured It Out

https://www.adama-platform.com/learn
2•mathgladiator•45m ago•0 comments

Show HN: Kubently – Debug Kubernetes Clusters Agentically

https://kubently.io/
1•drtydzzle•45m ago•0 comments

Show HN: I built a running coach that reacts to your training performance

https://strides.coach
1•garydevenay•46m ago•0 comments
Open in hackernews

The Generative Burrito Test

https://www.generativist.com/notes/2025/Nov/25/generative-burrito-test.html
56•pathdependent•1h ago

Comments

malkamius•1h ago
you be making me hungry
namegulf•1h ago
This is spooking our appetite
skocznymroczny•1h ago
That SD 1.5 picture doesn't look like base SD 1.5. It's way too good, perhaps it was some kind of finetune like RealisticVision?
pathdependent•51m ago
hrm. yea you're right. the page on fal used to produce it was linked with the image, but maybe i made a mistake and sloppily saved wrong one. ill have to reroll to check
ilaksh•1h ago
I'm so easily influenced. I came very close to immediately ordering Mexican food on DoorDash.
elzbardico•57m ago
Only nano banana looks somewhat partially-eaten.
adammarples•57m ago
Nano banana is incredible. What is their secret sauce?
jfim•51m ago
A training corpus that includes the images from Google image search probably helps a lot.
kbenson•56m ago
Oh wow, I've been hearing about Nano Banana Pro in random stuff lately, but as a layman the difference is stark. It's the only one that actually looks like a partially eaten burrito at all to me. The others all look like staged marketing fake food, if I'm being generous (only a few actually approach that, most just look wrong).
kemayo•50m ago
Hunyuan V3 is the only other one that plausibly has a bite taken. The weirdness of the fillings being decoratively sprinkled on top of it does rather count against it, though.
andai•29m ago
Hide the evidence!
Aloisius•12m ago
The NBP looks like a mock of food to me - the unwrapped burrito on a single piece of intact tinfoil, a table where the grain goes all wonky, an almost pastry looking tortilla, hyperrealistic beans and there's something wrong with the focal plane.

It's just not as plasticy and oversaturated as the others.

BoorishBears•2m ago
This shows some gaps in the "same prompt to every model" approach to benchmarking models.

I get that it's allows ensuring you're testing the model capabilities vs prompts, but most models are being post-trained with very different formats of prompting.

I use Seedream in production so I was a little suspicious of the gap: I passed Bytedance's official prompting guide, OPs prompt, and your feedback to Claude Opus 4.5 and got this prompt to create a new image:

> A partially eaten chicken burrito with a bite taken out, revealing the fillings inside: shredded cheese, sour cream, guacamole, shredded lettuce, salsa, and pinto beans all visible in the cross-section of the burrito. Flour tortilla with grill marks. Taken with a cheap Android phone camera under harsh cafeteria lighting. Compostable paper plate, plastic fork, messy table. Casual unedited snapshot, slightly overexposed, flat colors.

Then I generated with n=4 and the 'standard' prompt expansion setting for Seedream 4.0 Text To Image:

https://imgur.com/a/lxKyvlm

They're still not perfect (it's not adhering to the fillings being inside for example) but it's massively better than OP's result

Shows that a) random chance plays a big part, so you want more than 1 sample and b) you don't have to "cheat" by spending massive amounts of time hand-iterating on a single prompt either to get a better result

koakuma-chan•52m ago
Impressive partially eaten burrito by NB Pro
N_Lens•51m ago
Nano b̶a̶n̶a̶n̶a̶ burrito
drob518•51m ago
The burrito benchmark is poised to become an industry standard.
_joel•41m ago
Ricing a bit more performance out.
visioninmyblood•50m ago
Would be great to see video results for this as well. I generated some with other models. Nano pro seems the best so far
JumpCrisscross•47m ago
An interesting American culinary divide is between Scottsdale and Phoenix homemade burritos. The former being closest related to the Midwest variety, the latter to a Sonoran style.

Even ignoring the Heinz bean outliers, these are all decidedly Scottsdale. With one exception. All hail Nano Banana.

throwup238•13m ago
They all just look like generic Mission burritos to me (leaning towards fast food menu photos), except some include lettuce and some have blisters sonoran style. Only Nano Banana really looks like something I'd get at El Farolito.
corpMaverick•38m ago
I am disappointed there were not donkeys in any image.
jasonthorsness•35m ago
This progress bodes well for my chances of visualizing an invention I have been working on, a perpetual burrito extruding machine
ruined•26m ago
let me know when you're in preseed
minimaxir•18m ago
One of my tests for new image generation models is professional food photography, particularly in cases where the food has constraints, such as "a peanut butter and jelly sandwich in the shape of a Rubik’s cube" (blog post from 2022 for DALL-E 2: https://minimaxir.com/2022/07/food-photography-ai/ )

For some reason ever since DALL-E 2, all food models seem to generate obviously fake food and/or misinterpret the fun constraints...until Nano Banana. Now I can generate fractal Sierpiński triangle peanut butter and jelly sandwiches.

willio58•17m ago
I like how a couple of these basically show the model is confused between pinto beans and baked beans.
basket_horse•13m ago
Is no one going to mention fast Lightning’s sploogerito
totetsu•7m ago
With llms there is a secondary training step to turn a foundational model into a chat bot. Is these something similar going on with these image generation models, that is making them all tend towards making pretty clean images and stopping them making half eaten food even if they have the capabilities?
digitcatphd•5m ago
I find it a bit surprising GenAI has made it this far without this benchmark