Moonlit Monkey (@MoonlitMonkey69) header
Moonlit Monkey (@MoonlitMonkey69) profile picture

Kiwi. Pragmatic realist-centrist. Reason serves instinct, does not rule it. 'debate' w/o good faith, honesty, manners, or substance = blocked

The Circus
Joined February 2021
'the bitter lesson' is wrong. Reality is just far too complex to problem search to the point of emergent world modelling, at any reasonable scale. For scale to work well, it needs to not just be bounded, but already naturally compositional, rule-like.
1
3
255
First all the people hiding in military bases after 'threats', and now this 'dirty bomb' stuff. They are itching to get back to their pointless war.
New US intelligence indicates that Iran is developing plans to target senior Trump administration officials by detonating a so-called dirty bomb or a chemical weapon in Washington, DC, even as officials continue to debate whether Iran can or will attempt to execute such a plot, according to seven US officials. cnn.it/4903mvK
24
Oh boy, netflix has it's SBF miniseries 'The Altruists' ready soon.
1
2
109
1.9B worth of BTC ETF sales btw. More ETF sales than leverage flushed.
1
74
"We also plan to release Hy4 flash as an open-weight model" Thank the stars. Need someone to take up the 'can run it locally mantel' deekseek is slowly abandoning, along side qwen. hy.tencent.ai/research/hy4-f…
1
2
150
The only AI benchmark competition that matters to me: Models under 700B parameters that you could plausibly run local.
2
7
208
Most of these are honestly probably still too large outside of tiny quants with short context. But I'm not a renter. I like to own things.
2
3
37
Super interested to see how Qwen4 does here.
1
37
I like how after you fast for awhile, and then eat, your shit comes out like a water hose.
2
6
170
This price efficiency is pretty promising for their smaller open models.
📢 Meet Qwen4-Max, your new AI teammate for work, creativity, and discovery! Qwen4-Max is built on a new MoE architecture with native vision and a 1M-token context window. Highlights: - Outperforms Qwen3.8-Max with up to ~50% fewer tokens, while average cost per task drops over 60% in our internal evaluation. - Advances real-world productivity through greater proficiency with professional tools, stronger long-horizon execution, and higher-quality deliverables. - Handles real tasks, including 3D modeling, game development, manufacturing, finance, law, healthcare, and frontier scientific discovery. - $1.5/M tokens for input, $4.5/M tokens for output, and $0.075/M tokens for implicit caching — all lower than Qwen3.8-Max, with implicit caching down by 70%. The API is live on QwenCloud. You can also try Qwen4-Max on Qwen Studio, QwenWork and Qoder. Start building today! 🚀 - Blog: qwen.ai/blog?id=qwen4-max - API: qwencloud.com/models/qwen4-m… - Token Plan: qwencloud.com/pricing/token-… More to come! The open weights of smaller Qwen4 models will be released. Stay tuned. 👀
1
86
Relative to other competitors, the bench improvements are not as strong as prior versions. But this reminds me of the last new DS version increment: primarily focusing on efficiency first. We'll probably see stronger bench improvements later.
19
Is it normal for US people to call their strange private healthcare subsidy system 'public healthcare'?
94
Either a bottom forming, or consolidation before another lower leg. 4hr btc. Bulls have to hold this line, on corn.
2
3
173
Hopium wise, there is this (DXY), and makes a fairly compelling argument that the low might be in.
21
Crypto just giving up into a 1.5b liquidation cascade
3
217
e/acc is now just devolving into copy pasting bitcoin maxis.
Replying to @bofrench
It’s funny that all the racists on X have been arguing about which race is better while the entire world has been getting wildly wealthy during the AI boom. Keep distracting yourself and stay poor.
1
2
280
Top five things for no reason, and because no one asked (and not in order because that's too hard). Movies: The Usual Suspects Shawshank Redemption The Arrival Bladerunner 2049 Princess Bride TV: Devs Westworld Game of Thrones Andor Fringe
10
1
23
59,405
Games: Disco Elysium Pillars of Eternity 2: Deadfire Pathfinder: Wrath of the righteous Zero Parades for Dead Spies Arcanum: Of Steamworks and Magick Obscura
1
2
995
I ended up with all RPGs of some flavor, but I don't mind me a Fallen Order, or Tomb Raider type of game.
519
One has never been able to tell bullish structure from bearish structure scenario in this leg down. But a little more decisive now. 64k is kind of the last line in the sand for bulls (the .786)
1
2
155
Only real hope for bulls rn, is that volume is falling on decline. Which isn't necessarily bullish, but indicates more uncertainty from both sides of the trade.
1
50
IMO, minute we get a daily close below 64k (the .786, about as deep as a retracement should ever go), then the prospect looks immediately bearish. So far, so good tho.
1
1
20
ETFs, probably.
Update: BTC sell pressure continues to come primarily from the U.S.
85