Dude I asked it simply who the previous 10 nba finals matchups were and it couldn't do that
98.2% of Qwen3.8 27B my ass.
I got hyped and gave it a real agent job right away. Build me an FPS in three.js, 6 hours on a 3090. My most standard and default prompt that I always use. It spent the first 32K tokens on a plan without writing a single file, then shipped a black screen, 2 shaders that dont compile and a player who spawns dead. And wrote "verified" in the final report.
Ok, too hard. I gave it the easiest thing I have, a voxel pagoda garden in one html file. Video attached. 3 hours for THIS. On the way it deleted its own file and spent an hour debugging a raycaster nobody asked for.
Where the 98.2% is in all this I have no fucking idea. Same Qwen3.8 27B, same pagoda task, ISTA-DASLab GSQ-RCO IQ2_XS on a 12 GB 3080 Ti gave me day and night, real shadows and koi fish in the pond. 8.4 GB on disk, real 2.50 bpw, 131072 ctx with q4_0 KV, 47 tok/s at 128K. 2.5 bits beats 2.13 bits by a lot when the 2.13 is this shit. Post with the video and the exact command in the replies.
In the replies, I'll attach what a proper Qwen3.8 27B created in my hands. Bottom line: if you have at least 12GB VRAM, use ISTA-DASLab GSQ-RCO of all THIS.
1
2
329
Agent Heights retweeted
very excited for the future of @theprophetfun!! much more to come we got big plans for the future of agentic predictions
bare w me on a quick appreciate post this my first one lol
first off this project was a direct culmination of getting more involved in the @solana eco. we've had the idea of a prediction agent for awhile at @TradeLeagueHQ
but then I got to @mtndao and everything changed - I was exposed to all the different ways founders are building and growing their projects - and that's where we first started thinking about building this out
there are plenty of AI forecasters and even a few that are tokenized - but what if there was one that could learn from the crowd and compile their collective insights into trade theses? what if it could actually execute these theses? what if you could allow retail access to those trades and even execute pm trades on their behalf? what if you could have more than a stake in the agent and actually use the token to steer it's decisions?
3 weeks later @theprophetfun was born. huge s/o to @mtndao for providing the melting pot to mold this idea. much love to the @SP3NDdotshop boys @jpsychedelic369 @kenton_cooley for inspiring us to do this and @agentheights for giving us the confidence to pull the trigger! and ofc none of this would be possible wo/ @HajiMagege who went from zero to MVP in less than 10 days 💪
we're just getting started got a lot more to build! oh and if you think we forgot about @TradeLeagueHQ you're trippin
Week 1 Update 🧙♂️
Launched solana:LekVijLi3RFGWcsLFzcWAiDQjFxqcEZ21LDpGu22mhx on @clawpumptech and entered the @blknoiz06 hackathon 🚀
Alpha is live, dev supply locked, agent wallets built. Ready to prophesize 🪄
Full update below 👇
12
8
40
1,962
Using this now its pretty legit
We built the cheapest AI Router
Every model is discounted up to 81%!
ZDR is available
Try it -> CheaperInference.com
1
138
Agent Heights retweeted
"an infinite slop machine, a portable casino, and a 24/7 surveillance device."
"an infinite slop machine, a portable casino, and a 24/7 surveillance device."
"are you getting it?"
75
1,214
17,053
341,110
Agent Heights retweeted
When my IRS agent comes to visit me and asks' about my Monero:
$XMR
26
54
618
49,563
If an agent falls in the forest and nobody hears it, does it make a sound? or does he just pop back into the office?
1
1
193
*ahem
How is opencode so bad. Who is making a:
- polished desktop / mobile app
- quality model harness
- remote-first (runs on servers)
- able to orchestrate massive swarms
- uses open source models (third party inference)
Open source models are getting so cheap, now we can all turn our eye of stargate onto tough problems and let 100 agents rip for $50.
Will I have to make this?
3
322