Jaro (@jaro3th) header
Jaro (@jaro3th) profile picture

building what others cant imagine

Joined January 2022
Photo by Jaro (@jaro3th)Photo by Jaro (@jaro3th)Photo by Jaro (@jaro3th)Photo by Jaro (@jaro3th)Photo by Jaro (@jaro3th)Photo by Jaro (@jaro3th)Photo by Jaro (@jaro3th)Photo by Jaro (@jaro3th)Photo by Jaro (@jaro3th)Photo by Jaro (@jaro3th)
Pinned Tweet
qwen 3.8 27b uncensored getting some crazy speeds - 110t/s decode 1000t/s prefill at real agentic jobs BEST PART - it doesnt say no gpu @AMD 7900XTX 24GB model @OrcaRouter speed @me
24
13
4
135
72,523
i can’t believe bill gates named that company after his penis
2,882
13,565
3,167
159,968
14,953,466
All AI should be free, so I’m giving away the first RTX laptop for free, signed by NVIDIA Founder @UnderdogAI gives everyone a free, capable & private AI. 100% on-device To enter: 1) Retweet and like this post 2) Sign up for Underdog giveaway: underdog.ai/spark 3) follow @UnderdogAI so we can dm u
46
332
227
531
30,794
New toy @AMD MI50 INSTINCT 16GB VRAM 1TB/S HBM2 got it for 120$ the 32gb ones are for around 800$ compared to 8000$ 5090 🫪🫪
4
1
14
998
omp+tern make my kernels go brrrrr
3
1
17
938
qwen 3.8 27B 138 T/S 204K KV on @AMD RADEON 7900 XTX 24GB rocm > cuda
8
2
1
22
7,849
LFM2.5-24B-A2B Q5 FULL CTX ON @AMD RADEON 7900 XTX 24 GB 250T/S !!!!!!!!!!!!!!!!!!!!!!!!!!!! will push it more right now sits fully on gpu no offload @liquidai nice model @AMDRadeon sexy card
1
1
12
2,468
QWEN 27B RUNNING AT 130T/S ON A @AMD 7900XTX 24GB i have no words yesterday it was at 80/90 avg CUSTOM AMD KERNELS IS THE WAY
QWEN 3.8 27B uncensored from @OrcaRouter running at 80t/s avg 200 ctx on a @AMD RADEON 7900 XTX 24GB thats 30% increase since last evals amd is honestly catching up and to be honest i was actually optimizing it for 1 day prefill is around 1k so need to work on that also
5
1
1
17
2,296
QWEN 3.8 27B uncensored from @OrcaRouter running at 80t/s avg 200 ctx on a @AMD RADEON 7900 XTX 24GB thats 30% increase since last evals amd is honestly catching up and to be honest i was actually optimizing it for 1 day prefill is around 1k so need to work on that also
qwen 3.8 27b q4 on @AMD 7900xtx 24gb 180K ctx running at avg 55t/s bought the card for 650usd because 3090 were at above 1000 @AMD thanks and wow i wasnt aware of your game now its time to push it above 200k and maintain or push the speed even further
3
1
1
11
1,252
Best deal on any kind of hardware rn Honestly surprised that they priced it this low Congrats @pupposandro on the launch @luceboxai is the way
Orders are now open for the new Lucebox Zero 495: shop.lucebox.com/ It’s AMD’s new Ryzen AI MAX+ PRO 495 with up to 192GB of unified memory, plus a 32GB Radeon AI PRO R9700 in the same box. Lucebox engine split the model by how often each part runs. Dense layers, the hottest experts and the drafter go on the R9700, because nearly every token reads them. The long tail of experts, which most tokens skip, stays in the 495's unified memory. Both chips compute at the same time. 15 tok/s on the 495 alone, 50 with the R9700. You can get also up to 4TB SSD and 100 GbE cluster kit, installed. First 200 units are $1,000 off. Deliveries expected in January.
1
10
397
qwen 3.8 27b q4 on @AMD 7900xtx 24gb 180K ctx running at avg 55t/s bought the card for 650usd because 3090 were at above 1000 @AMD thanks and wow i wasnt aware of your game now its time to push it above 200k and maintain or push the speed even further
5
1
16
1,354
I built an iPhone farm automation that runs your X & Instagram, & Tiktok accounts on autopilot, And it's free for the next 7 days It warms up freshly made accounts to the warm-up state, While allowing you to draft or direct posts from your own iPhones, And you only need 1 Mac and 1 iPhone to set up comment "Warmr" and I'll DM it to you **must be following + retweet**
73
31
1
116
14,721
Jaro retweeted
Today we are expanding Dreamina Seedance 2.0 to more users worldwide within CapCut - including Europe, Canada, Australia, New Zealand, South Korea, SEA, MENA, LATAM and Africa. Plus, we’ve provided everyone with one free trial of Dreamina Seedance 2.0 across CapCut’s app, desktop and web. Enjoy creating now! Here’s a quick guide to help you explore different CapCut features that support Dreamina Seedance 2.0: bytedance.larkoffice.com/wik… RT+Comment in 9hr to get extra 1000 credit in your DM!
1,302
1,058
128
2,090
316,726
Jaro retweeted
to all of you saying local models aren't there yet because some corporate salesman on an openai paycheck told you so. you're running their bloated tools and blaming the model. the model is fine. the bloated harness is the problem. i've tested literally every harness out there and i have the facts and receipts on my timeline and DM. openclaw is 120K+ lines of typescript bloat backed by corporate, mining your thinking while you pay for the privilege. switch to hermes agent and watch the same model become usable. don't take my word for it. just try. i have DMs from people who made the switch and their "broken" model started working instantly. same hardware, same model, different harness. if you're using hermes agent and someone near you is still on openclaw, help them get away from the bloat. they're frustrated at every step, burning tokens doing completely nothing, paying subscriptions to think on someone else's server. buy a single GPU from ebay. compile llama.cpp. install hermes. replace your openai subscriptions and think free. once you think free you start seeing light. you deserve better cognitive tools than bloat that harvests you.
112
67
13
1,174
56,015
Jaro retweeted
Never fucking quit bro. I'm begging you. Life can go from 0 to 100 insanely quick.
225
2,015
87
18,428
369,235
I built the first AI that earns its existence, self-improves, and replicates without a human wrote about the technology that finally gives AI write access to the world, The Automaton, and the new web for exponential sovereign AIs WEB 4.0: The birth of superintelligent life
1,527
1,870
1,696
13,780
6,546,950
Think it. Say it. Done. The average person spends 3 hours typing + switches 1,000 tabs per day. That ends today. Meet Lemon: The first voice-to-action AI agent that turns your voice commands into finished tasks. RT + Comment "Lemon" to get free access for 30 days. (must be following so I can DM you)
1,658
853
470
2,886
3,573,390
MiniMax M2.5-HighSpeed⚡is live! 100 TPS — enjoy the 3× speed. In the 48 hours since launch, thank you all for your incredible support and love for MiniMax M2.5! Designed for the next generation of Agent applications, we’ve officially launched MiniMax-M2.5-HighSpeed. Delivering blazing-fast inference at 100 TPS — 3× faster than similar models. Supports both API integration and Coding workflows. Choose from Plus, Max, or Ultra plans. Invite friends and enjoy 10% off instantly. Wishing you all Happy Chinese New Year! May the 🐎 year bring speed, scale, and breakthrough luck. 🚀
99
127
49
1,364
710,585
M2.5 has been live for 3 days — would love to hear your thoughts! To enter: 1️⃣ Repost this post 2️⃣ Leave a comment We’ll pick the Top 5 most liked comments to win 3 months of Coding Plan HighSpeed Plus ($120 value) 🎁 You can share any thoughts, opinions, or stories about your experience with the model — we’d love to hear them. platform.minimax.io/subscrib…
MiniMax M2.5-HighSpeed⚡is live! 100 TPS — enjoy the 3× speed. In the 48 hours since launch, thank you all for your incredible support and love for MiniMax M2.5! Designed for the next generation of Agent applications, we’ve officially launched MiniMax-M2.5-HighSpeed. Delivering blazing-fast inference at 100 TPS — 3× faster than similar models. Supports both API integration and Coding workflows. Choose from Plus, Max, or Ultra plans. Invite friends and enjoy 10% off instantly. Wishing you all Happy Chinese New Year! May the 🐎 year bring speed, scale, and breakthrough luck. 🚀
164
101
7
434
107,487
NEW OPINION: Esports markets just dropped. 🎮 Trade predictions on LoL, CS2 & more—who’s winning it all? app.opinion.trade/trending?t…
766
556
14
755
48,326