79 comments

[ 7.3 ms ] story [ 144 ms ] thread
Should I be able to see the actual results? Because I don’t see a hangar anywhere.
Click Open in the file column
File -> Open

If you don't have anything working check the console, maybe a WebGL issue.

Qwen with open code seemed like the best to me. It seemed like the best balance between performance and visuals
Astra has so much more detail. It did take 5x more time, but that's usually negligible when it replaces x hours of human output.
It’s really interesting how much little choices make the result better or worse. Astra and one of the GLMs added bright lights, and thus looked so much better to my eye.

Genuinely happy with some of the Qwen 3.8 results (especially since I can run that model at Q8).

Interesting to see how much better (at this task) Pi (OMP) is over Opencode as a harness.

I’d love to see a few more with outcomes that are as easy to judge but less subjective.

I’ve got a toy project going to make a fun to watch battle simulator where an LLM (or two if playing vs) has to write programs that control multiple bots (each with their own line of sight and limited battle context) that have to coordinate and fight alongside each other. Goal is to have the LLM update the code based on current situations maybe 5-10 times in a 5 min simulated battle. Exploring even allow the bots to request new programming and score based on number of reprogram steps.

Interesting also that Astra and Sol didn't check visually using the browser. I suspect they would have done even better if they had.
huh. goes to show how much browser selection impacts things. the OMP Qwen result is completely unresponsive in free cam on my iphone, whereas the opencode version is responsive, has multitouch zoom and multitouch pan. the glm models do really well with this as well, but all of the anthropic models have pretty gimped camera controls, even astra, which while pretty wont let me pan and only allows me to rotate about 120 degrees! I’m really impressed with the Qwen 27b OpenCode result.

I suppose the only thing that the prompt asks for is the cinematic view, and honestly they all kinda fail on the “subtle volumetric-style fog planes”, none of them have more fog when you get farther from a light source.

Yeah, most of them made it too dark to see what's going on.

Also your last paragraph sounds like the setup for a late 80s scifi movie...

I wish there was another column with the estimated cost for each, with a specific date.

Ideally also finding somehow (not sure what would be the right away) what is publicly available before running the test. It's quite a different outcome if there are competitions, e.g. js13k, live code examples from books, even templates, on specific that topic. Visually here the results looks very very similar to the point that I can't help but wonder if it's the result from the short yet relatively descriptive prompt or because some template was always found and relied on.

Nice. But these tests raise a question what results are reproducible, the final visual, time, tool calling or it's mostly noise. Like, have you tried same combination or model and harness multiple times?
I have on Qwen3.8 27B. Results are close enough across multiple runs.
So it's not reproducible.
This is inevitable unfortunately with LLMs.
What I'm missing in all of those one-shot tests:

How different are the results between multiple runs of the same setup?

I have done multiple runs on Qwen3.8 27B. Results are close enough across multiple runs qualitatively.
Refreshing to see a post that gets straight to the point and not hide the data behind pages of prose.
The Astra version seems to have used three.js r170, which is from October 2024. Sol used an even earlier version. GLM's code used the latest version, but I think it's just getting three.js@latest from jsdelivr so it's unlikely to be writing code against that version. Qwen on OpenCode also fetches from jsdelivr, but using a pinned version at r160.

I don't think any of these examples are using things like tone mapping so they're stuck in sRGB (AgX or ACES look much better), they're not using the node materials (good for programmatic texture implementation), and they're not doing anything cool like baking shadow environments or using post-processing effect.

They're nice, but I think they're showing how far behind AI models are on this sort of project rather than how good they are.

Really appreciate this perspective. It’s really tempting to only be amazed at what these models can do, but my experience matches yours, they need help with details to do well.
What’s a good way to drive models to use current versions?
Anyone know of pertinent skill files to nudge the agents into these directions?
Don't use skills. Point it to a game engine or better yet full game repo or whatever and ask for what you want.
Not tone mapping/shader graph is probably the least of the sins, these kinds of one-shots tend to produce hundreds of THREE.Geometry, massive matrix walks, etc. that explode once you move beyond a screen saver.

The models _can_ do it, but you need to ask for the right things. Most people don't, they'll usually blame the browser for being slow or ugly when they can't break through the THREE demo page wall.

The fact that these tools are capable of doing anything at all is a miracle to me.
How am I driving? How does an engine even work?
That's a very strange position.

There are thousands, if not millions now, of people working on the problem, a vast majority of humanity text output used for training, an enormous infrastructure composed of the most complex human made device (the processor), a gigantic amount of energy running it, for an attempt that spanned decades, if not century, heck if not millennia if you go as far as considering Antikythera.

It might be impressive but it's anything but a miracle when we consider how much effort was poured into this.

I love everything about OpenCode except for the stuff it outputs. On paper it has everything I want in a harness and more, but it tends to struggle to deliver the desired outcome. Last I used it the context was massive and tool calls were reliably unreliable. Is it worth revisiting as a daily driver?
Why not ohmypi?
Cache rate for oh-my-pi is significantly lower than opencode's in your results.
i moved from opencode to omp and i don't think I'm ever going back
Btw in my quest for a good desktop codex/claude like app - https://github.com/openchamber/openchamber - this seems to be front-runner. I pair it with OMP via https://github.com/alvins82/omp-openchamber-server/.

I wanted a powerful GUI+harness setup for open models so I could use/test as they came out.

I like Paseo https://github.com/getpaseo/paseo

But I am annoyed at these GUIs implementing features I don’t care about. I want them to just wrap my harness and forward it to my iPhone, but they can’t help themselves from feature creep.

I've been impressed with orca

https://github.com/stablyai/orca

I did have some issues getting it installed on a headless server. I sort of gave up and installed the instance that I use as the remote server on a Debian + xfce machine I had laying around.

I found it appalling when working with remote setup — tabs disappearing and appearing after some time, screen hangs up, projects disappear. Tried to fix some issues with PR, but team is not accepting for 2w+.
i liked paseo too but its not optimized at all. let alone the feature bloat.
What do you use then?
So where did most of them get the idea of calling it hanger 07?
Useful comparison. One thing a single-shot, single-file test can't show is how much of the difference is the harness rather than the model. In day to day use the same model behaves pretty differently between harnesses on an existing codebase - tool calling style, how much context it pulls in, whether it verifies its own edits. Would be curious to see this matrix on a task that modifies a multi-file project instead of generating one self-contained HTML. Greenfield single-file output is close to what these models see in training, so it tends to flatter all of them.
What is DSH / PTC?
Deepseek Harness. PTC is one of the 'modes' it has.
Ptc is programmatic tool calling or code mode

Basically it can run these mini programs where each input might be another toolcall, so it can run without waiting for whole LLM response and ready the parameters async.

I’m always confused, are all these shapes programmatically generated or are they downloaded from some source?

Also I think Astra looks the best and has the best functionality. Also shocked how much better GLM is on the Non Codex harnesses. Didn’t think it would make such a difference.

Would be nice if you could include cost in the table

I saw examples on X of people building amazing things with Three.js and models, and I figured out how to build them myself. But the results are mostly low-poly materials and crappy animations. I thought it was my prompt (it was detailed), but it seems to be mostly a limitation of the models.

From all the examples I've seen, Astra does it really well, and I suspect it's because they wanted to attract game designers, so they trained the model more on 3D, animation libraries, etc.

Sol and Opus are pretty good with the right prompting and feedback framework.

Or, my standards are lower. It's sometimes hard to tell in these discussions whether people are talking about getting production quality results, or stuff that's good enough for a one off blog post.

Would be good to include Fable and Opus. Based on the results with Astra vs GLM, these are probably the realistic alternatives to consider.
Yes, I'd be really interested to see Fable / Opus / Sonnet.

GLM 5.3 Flash Max had an interesting showing. Its Codex version was bad [0], and it completed in 9 minutes. The OpenCode version was much richer [1] and more detailed, completed in 20 minutes. And the OMP version was arguably the most complete [2], completing in 30 minutes.

This is probably the strongest argument for the effect of a harness, and I'd be interested to learn the differences in the prompts and tools between these three.

[0]: https://alvins82.github.io/hangar-harness-model-tests/hangar...

[1]: https://alvins82.github.io/hangar-harness-model-tests/hangar...

[2]: https://alvins82.github.io/hangar-harness-model-tests/hangar...

(comment deleted)
This is rad, the Sol one is probably the coolest.
to me i think the diffrenece is not that big?
fun benchmark. did any harness consistently win regardless of model, or is it more about the pairing?
Am I missing something or does sorting by Duration not work? Chrome on Android mobile