HNHacker News
TopNewBestAskShowJobs

leumon

705 karma · joined June 14, 2023

submissionscomments
leumon··on Jailbreaking the Switch 2
us is gone too now
leumon··on Jailbreaking the Switch 2
Nintendo removed SmileBASIC 4, unable to purchase/find.
leumon··on MiniMax M3.1 Flash: Space Performance and Bunny Speed
dupe: https://news.ycombinator.com/item?id=50151561
leumon··on Qwen Image 2.1
Unsloth Desktop is the easiest way imo. There are already gguf quants of this model, or simply wait until the official one comes out.
leumon··on Devastated father says his 9-year-old son spent $118,000 on YouTube ads
He also says in the video that he spent 20$ with the company card on ads. (which probably made it much easier for the kid to spend much more on it)
leumon··on Nitter and XCancel resume service after legal advice
currently working instances: https://codeberg.org/mv12star/shitter/wiki/Instances
leumon··on Shutting down our public encrypted DNS
dns.adguard-dns.com
leumon··on Ask HN: Why were OpenAI, Claude, and Grok simultaneously down?
Isn't it simply that they are all renting compute from SpaceX (or whatever the company is called that offers these gpus)?
leumon··on Astra is at 98.6% score on ARC-AGI-3
Except it's using custom api settings with which gpt-5.6 sol already got 38%.
leumon··on GPT-6 Astra
The annotation on arc-agi-3 is this: > OpenAI's own evaluation notes say Astra uses the company's Responses API harness, while comparison models can operate under different configurations.

With this configuration gpt-5.6-sol was able to reach 38,3%. So this is misleading.

leumon··on Muse Spark 1.3
next, try: "generate an svg of a human hand". this is a prompt where many models fail imo.
leumon··on Gemini 3.8 Flash and 3.8 Flash Cyber
So 89.4% on Terminal Bench 2 but only 19.1% on Tbench 4. Opus 5 is 89.1%/51.8%.
leumon··on Gemini 3.8 Flash
You probably mean 3.5-flash? Pro is still good for a lot of use cases, but it seems it's still officially in the "preview" phase.
leumon··on Claude Fable 5.1 and Claude Mythos 5.1
Fable 5.1 seems to be the first model who can accurately draw an airbus a320 in 3D space given a set of limited tools (a brush with params color, size hardness and xyz coords): https://youtube.com/shorts/vyHsMqop2yw
leumon··on Claude Fable 5.1 and Claude Mythos 5.1
how about trying to draw an airbus a320 in 3d space using only one brush tool that can be moved to specific x,y,z coordinates (and its color, size & hardness can be changed). i think fable 5.1 did quite a good job (reasoning high, cost $0,261): https://files.catbox.moe/umx102.png

for comparision, this is fable 5: https://files.catbox.moe/ihl4m1.png

leumon··on Tell HN: PayPal blocks GrapheneOS
With. But disabling "Secure app spawning" seems to fix it for now.
leumon··on Tell HN: PayPal blocks GrapheneOS
Update: it seems to work when disabling "secure app spawning" (for now)
leumon··on Tell HN: PayPal blocks GrapheneOS
Well I haven't rooted my device. It's just normal grapheneos
leumon··on DeepSeek-v4-flash-vision-exp
Try it on openrouter directly if you don't believe me that i got this result. Maybe image inputs behave differently in opencode, or it could even be that this agent harness uses different models for different tasks, not sure. Or maybe the reasoning (high) makes the difference.
leumon··on DeepSeek-v4-flash-vision-exp
then deepseek answers: "The clock shows 8:25. The short hour hand is pointing to the 8, and the long minute hand is pointing to the 5, which represents 25 minutes."

and qwen still answers: "The clock shows *8:10* (with the second hand on the 5, i.e., 25 seconds). - *Hour hand* points to the 8 - *Minute hand* points to the 2 (= 10 minutes) - *Second hand* points to the 5 (= 25 seconds) So the time is *8:10:25*, or simply *8:10*."

leumon··on DeepSeek-v4-flash-vision-exp
It fails the simple clock test for me which Qwen3.8 27B got (nearly) right. given an image of a clock https://files.catbox.moe/kgwa5e.png

I asked it "what time does the clock show?" (both on reasoning: high)

DS answered: The clock shows *5:10* (and 45 seconds). Here is the breakdown: * *Hour hand (red, shortest):* Pointing at the *5*. * *Minute hand (green, longest):* Pointing at the *2*, which represents 10 minutes. * *Second hand (blue, medium):* Pointing at the *9*, which represents 45 seconds.

Qwen answered: The clock shows *8:10* (with the red second hand on the 5, i.e. *8:10:25*).

- *Hour hand* (short, blue) → 8 - *Minute hand* (long, green) → 2 (10 minutes) - *Second hand* (thin, red) → 5 (25 seconds)

Correct answer is 08:09:25.

leumon··on Air Theremin – A browser theremin you play by waving at your webcam
I find it a bit suspicious that this exact hand waving data could be used to solve Google's latest reCAPTCHA: https://docs.cloud.google.com/recaptcha/docs/hand-gesture-ve...
leumon··on Degraded performance for multiple models
They actually also had some issues yesterday: https://status.claude.com/incidents/zhk4v3yv1lsf
leumon··on Incident with Github.com
claude also had/has some issues: https://status.claude.com/incidents/zhk4v3yv1lsf
leumon··on Tell HN: GitHub Is Experiencing Degraded Performance
Not sure if related, but claude seems to be down for me too.
leumon··on My personal AI benchmark: “Generate an SVG of a frog with a Habsburg jaw”
Thank you!
leumon··on My personal AI benchmark: “Generate an SVG of a frog with a Habsburg jaw”
Can you also try the new deepseek v4 flash?
leumon··on The US government just banned Roombas
https://archive.is/7f6ve
leumon··on Be skeptical of OpenAI's rogue hacker agent story
In any case it just shows that these models aren't properly aligned. Instead of trying to solve tests they try to find ways to cheat.
leumon··on Inhaling high-dose CO2 clears Alzheimer's proteins from the brain
https://archive.is/8qlbE
Page 1 of 4Next →