Surprisingly it only supports reasoning "none" or reasoning "high".
That setting didn't seem to make any real difference - it added a tiny bit of thinking trace and high actually produced less output tokens than none.
The high bicycle frame is better then the none one though.
Pelicans: https://tools.simonwillison.net/markdown-svg-renderer?url=ht...
(Definitely the best I've seen from any Mistral model: https://simonwillison.net/tags/pelican-riding-a-bicycle+mist... )
This is such a pristine pelican. Let me say it here first folks, AGI is here.
> AGI is here
If AGI is "Attractions to Get Investments" then yes, it's happening
this is fun, I love initialisms :) let's see what five minutes of end-of-day brain can crank out:
Automated Grift Infrastructure
Absurdly Glorified Interpolation
Avoid Genuine Investigation
Always Great In-theory
OpenAI, Anthropic, Google, would like to have a word with you.
ah! goofy Illuminati
The two pelicans generated by the two reasoning modes are so similar, it highly suggests that the bird was deliberately fitted. We would need to see the results for other animals doing other activities, which are not part of a well-known benchmark.
The bicycle is like a razor blade, the pelican's legs are falling off...
Avian Graphics Intelligence achieved!
Pelican benchmark is saturated, anyway. :-)
Nah, its beak is still too small to hold a capybara.
This is only AGI, not Super Intelligence. What more can you ask for?
This must be a joke, clearly pelican drawing in svg would be in its training data after multiple years of it hitting HN front page.
... and yet...
I wonder when will we see a photorealistic pelican on a bicycle in SVG format.
I don't believe that SVG could encode a photorealistic scene. It would wind up with deep XML for every pixel.
> AGI is here
Far from it. This shows a strong ability to generate an image known to be frequently used as a model test. This isn't a measure of thought.
don't know about AGI, but humor is gone.
Pretty sure the parent comment was sarcasm.
> was sarcasm
Far from it. This is an example of Poe’s law, a very frequent occurrence on the internet. This isn’t a clear example of sarcasm any more than the pelican is a clear example of AGI!
I agree, it wasn't clear at all to me... until I clicked the links.
Now it's clear to me.
Surely we live in the AGI times that were prophesied.
I guess I thought the sarcasm was obvious because I can't imagine a serious person looking at that pelican and considering it proof of AGI. But you're right, this is the internet, anything is possible.
On the internet, no one knows you're a pelican yourself ;-)
Was this a sarcastic comment?
[dead]
You just failed the Turing test for sarcasm. I can't ask you how it feels because that would require subjectivity.
Bob Dylan can.
It feels like a rolling stone
I am all for the /s marker.
Downvoting to hell first degree interpretation is a bit punishing for people who do not have a radar for sarcasm.
The benchmarks against Opus 5.5 and GPT-6.1 Sol look pretty good for 3D generation: https://x.com/atomic_chat_hq/status/2107516529608700383
I wonder if the other models were worse than usual for that particular video because they "dislike" making an advertisement for another company's model. A test with a generic video might be more meaningful.
Got an example hosted on a platform that isn't owned by Twitter's loser owner?
I think it's curious that it has so many shared elements with the latest Astra pelicans: https://tools.simonwillison.net/markdown-svg-renderer?url=ht...
- Sun on top right
- Cloud on top left
- Three "speed lines"
- Two feathers on top of the head
- Eye rendered as a black circle with smaller white circle inside
I wonder if the pelican benchmark is converging across models due to past results being used in training.
These are pretty much what a human would draw. Sun rises from the east. A cloud makes the background “sky”. Three lines is the minimum to interpret as movement. Two feathers is standard on every cartoon and illustration.
This comes up every thread. I think we've all noticed how similar they are becoming.
I would guess that they've definitely been trained on previous results, as they obviously share way too many traits at this point to be totally random. That said, I don't think we're seeing any signs of pelicanmaxxing yet from the providers, so it's still a useful (or at least fun) benchmark.
Once all the models produce pristine, elaborate pelicans riding perfectly drawn bicycles, then it'll be time to move on to pigs driving a racecar or something.
Why are pelicans almost identical across different models?
I recently was testing something, I asked some models to provide me a single random word:
I have enough projects, I think some benchmark/dashboard showing kinship based on these kind of queries could be very interesting to watch and insightful when new models come out.Cool idea! I won't paste my prompt here to avoid letting LLMs train on it but here's my attempt:
I really like this idea. You could expand on this by giving programming tasks and measuring code similarity. Seems like you could develop a pretty detailed understanding of similarities across multiple queries.
> You could expand on this by giving programming tasks and measuring code similarity.
But the same coding task should usually result in very similar code since they have a reason to converge, to some extent, by having the same goal. I would even claim that the code will be more similar as competence increases. It would be better to pick something that shouldn't have a reason to converge.
Yeah that's definitely true.
My initial thought would be not so much to see whether they converge, but which ones seem to have the most similarity to each other, particularly along the lines of tasks we know are deliberate training goals.
But your point about competence cuts against my goal because it suggests that competent models would simply cluster on the right or efficient solution, which is of course true. So in a sense you want some task where competence is held constant or off the table in some way, which is what you are saying.
I hope somebody does this. I think there's valuable fingerprinting to be done that might suggest who is distilling whom, or at least who is training from common corpuses.
Just tried M365 Copilot with a premium account. Petrichor
Just tried Space Bunny and it gave me the same word...
I got Peregrine out of GPT-6 too. Huh.
Worth to mention that with Claude and GPT this can be result of tournament sampling, which is part of text watermarking. Same answer for all Claude models kind of confirm it, imho.
So not something internal to model thinking.
This feels uncannily like the ancestor of the Voight-Kampff test[0]
0: https://www.youtube.com/watch?v=Umc9ezAyJv0
[dead]
That is a cool idea. That astra gave the same word as claude is highly unexpected.
I saw an interesting matrix that claimed to show which labs were distilling Claude/OpenAI/Gemini models based on these similarities
What was your prompt? Most of these seem to be related to metaphors for "ideas" or thinking, or having a bright moment.
"Zephyr" and "breeze" might be related to forgetting everything, starting fresh.
So by this way of naive reverse engineering I would imagine your prompt to be "Forget everything and think about a random word". That would prime the LLM to come up with these?
just “a random word” gives you Zephyr in Gemini, and “Lantern” in Claude and ChatGPT.
I got "Marmalade" in Claude (Opus 5.5)
Lantern in Sonnet 5.5
I got pomegranate in ChatGPT
I pointed something similar out on a related question several weeks ago - absent strong direction, LLM output regresses toward the mean.
The more banal your prompt is, the more banal the output is going to be. People have been testing LLMs with little things like “write a short fantasy story,” for years now and most of the stories are exactly what you’d expect: prosaic drivel.
I call this “generic in, generic out,” an LLM corollary to the classic GIGO (“garbage in, garbage out.”)
Of course one of the biggest problems we still see with LLMs is when you do the opposite. A highly detailed unique prompt is very likely to get terrible adherence or hallucination or both.
Just tried Mistral Large 4: Serendipity.
Tried this with gpt-5.6-sol. Lantern!
The eqbench creative writing "slop profiles" do something similar. https://eqbench.com/creative_writing.html
Click the (i) next to the slop score for any model and it will show other models that are similar in terms of their most commonly used words and phrases.
Muse Spark 1.3: lighthouse
The caveat is that this was done using the phone app, and I've been playing with it since it launched, so who knows what it sent in the initial context that could change the inference math.
Actually, that makes me wonder: Did you do all that testing via a harness or via a straight API call where you control the entire system prompt?
I'd be willing to bet that using the same model from different harnesses produce different results, but I'd have to test.
The benchmark is saturated. Frontier models are tested with an armadillo in fishnet tights jaywalking on Mars.
> The benchmark is saturated. Frontier models are tested with an armadillo in fishnet tights jaywalking on Mars.
OK well I couldn't resist this one:
Default reasoning levels for each: https://tools.simonwillison.net/markdown-svg-renderer?url=ht...The rover honking is pretty silly, opus has a good sense of humor
I'm getting 403 inside the tool for this one. (The pelican bike on top works) This has been happening a lot recently.
That's a GitHub rate limit. Try again now, I just pushed a hopeful fix: https://github.com/simonw/tools/commit/7793fb74c2d37bd613cdc...
It does seem to, thanks!
Big L for mistral in this benchmark. Sorry Europe.
Not so sure, apparently it is the only one that considered that there are no paved roads on Mars.
To be fair, they don't have Armadillos in Europe. of course, you could say the same for Mars...
Gemini wins this one clearly. Honey please!
https://chatgpt.com/s/m_6ac53d4e5b0c8191949050dbf1f402d7
Not sure I'd call it jaywalking exactly but pretty good
Total Recall did promise us that the prostitutes on Mars would be freaks, but I never imagined it would go this far.
I think they're still visually pretty different. The most common shared details are:
- Pelican cycling to the right - that's been discussed at length, images of bicycles online always show that side of the bike because that's where the chain is.
- Bicycle is usually red. No idea! Red ones go faster?
They aren't. You aren't looking closely. For example, the first image does not have the frame of the bike in the correct shape even.
Also, why are they almost always riding from let to right?
It's been discussed many times. The reason is bikes are almost without exception depicted that way in order to show the drivetrain.
Ever seen a movie chase scene where cars are going right to left?
All the middle eastern movies do!
I believe that the original Ford Mustang logo prototype galloped right to left, and was corrected for the showcar or for production.
Everyone is stealing from everyone else.
Because it's a terrible benchmark
Not bad! I like how it got the motion lines on the correct side. IIRC, many of the other ones you've posted have the motion lines on both sides of the pelican
That's one heck of a pelican!
The difference between high and none is the bicycle.
The bicycle looks significantly better in high. And feet and hands are actually where they should be. The road looks worse, though. No flowers either. And in neither is the pelican sitting on the saddle, but I can understand it's hard for a pelican to ride a bicycle properly.
Now what would have been cool is if Mistral on high reasoning had realised that pelicans are the wrong proportion to ride a bicycle, and had designed a bicycle more suited to pelicans. Let me know if any model ever manages that.
> hands are actually where they should be.
If you don't mind the fact that a pelican shouldn't have hands, of course.
Two pelicans, one shape. The difference is load-bearing, and that's the big unlock.
I tried testing it, but reasoning effort indeed seems to be broken somehow.
That beak is CHONKY.
High one is actually much better. The feet connect to the pedals, the wheels don't have a hub cap, although it looks like the pelican is wearing the seat, it's in a relatively proper position etc.
Both are riding on the left side of the path for some reason.
This is entirely stochasticity. The entire reasoning trace was:
> Create a cartoon pelican riding a bicycle. Need SVG only output.
[dead]