Super excited for our open-weights release on Thursday... A new LLM that is almost FREE but massively improves on long-running personal agentic loops Yes, it will be available via the API and be better than DeepSeek Flash. Sneak peek at our NYC conference tomorrow! 🚀

ALT Season 1 Fire GIF by Game of Thrones

Sep 9, 2026 · 3:38 AM UTC

99
88
2,782
4,024,525
Sort replies: Relevant Recent Liked
Replying to @bindureddy
Open weights plus long-running agents could be a huge leap
12
Replying to @bindureddy
better than deep seek flash im giving a 0% chance here
1
16
3,171
Replying to @bindureddy
Does it think in latent space?
3
1,923
Replying to @bindureddy

ALT game of thrones television GIF
2
215
Replying to @bindureddy
A new open-weights model wouldn't be a bad idea need to compete with the open Chinese models
2
1,077
Replying to @bindureddy
Can it reach into other spatial dimensions and stydy consciousness yet? You should build it with biologics - the little lab grown brains. However, i do foresee a matrix scenario here....
2
1,999
Replying to @bindureddy
The old flash or the new flash ?
2
1,969
Replying to @bindureddy
Looking forward to testing what it can do. I only use the current version for simple tasks, but its name deserves so much more 🐉🔥
1
1,154
Replying to @bindureddy
Is it better than GLM-5.3 ?
1
1,650
Replying to @bindureddy
Almost free and better at long agent loops than Flash? That’s the combo that actually matters. Can’t wait for the weights
8
Replying to @bindureddy
Almost FREE and better than DeepSeek Flash for long-running personal agentic loops? Can't wait for Thursday's open weights drop see you at NYC!
51
Replying to @bindureddy
The “almost free” part may actually be bigger than being better than DeepSeek Flash. If inference gets this cheap, small businesses can start running agents for things like customer follow ups, quotation analysis and routine monitoring without worrying about the API bill.
7
917
Replying to @bindureddy
சீவக சிந்தாமணி கேட்டு அதனின் பால் மருகிய சோழனை,அஃதை விட பெருமிதம் மிகு அடியார் பெருமக்கள் வாழ்ந்த நற்றமிழ் நாடு என்று பெரியபுராணம் ஓதித் தேற்றிய நாட்டினில்,EVR அவர்களின் கொள்கைக்கு இணங்காக"தமிழ்ச் சனியனை விட்டொழியவும்",தமிழை விட Bible சிறந்தது என்றுரையும் "பரங்கிய மத போதகர்கள்"🔥
3
52
Replying to @bindureddy
open weights plus near free is the combo that actually matters for agentic loops. running an agent for hours is what kills half my ideas on cost alone. one i can run locally that grinds all day is the dream. you at the nyc thing tomorrow?
1
3
884
Replying to @bindureddy
Almost-free open weights that can actually sit in a loop for hours is the only kind of launch I still get excited about. Everything else is just another chatbot drop.
2
683
Replying to @bindureddy
"Almost free" is doing a lot of work in that sentence 👀 — curious what the actual token pricing lands at once it's live. Anyone have early benchmarks on long-horizon agent tasks specifically, not just single-turn?
2
528
Replying to @bindureddy
I don't buy your marketing hype. If it actually outperforms, I'll come back and apologize.
1
494
Replying to @bindureddy
Actually running code that decides things by itself? Thursday, pull the prices from 4 places: Anthropic, OpenAI, DeepSeek Flash, and hers. How much you pay each time matters most. Do the math.
1
1,509
Replying to @bindureddy
dev life never clocks out
1
65
Replying to @bindureddy
what base model and what did the eval look like against it on the agentic loop tasks
802
Replying to @bindureddy
Good morning
1
21
Replying to @bindureddy
又来了，「几乎免费」这四个字最近比模型迭代还勤。真要卷个人 agent，别只晒 token 价，晒一下长任务跑完一次多少钱、翻车率多少。👀
1
409
Replying to @bindureddy
Lower costs make AI experimentation far more accessible to small businesses. The real value will come from applying agents to specific repetitive problems rather than adopting them simply because they are available.
1
1,065
Replying to @bindureddy
Let's see what you are offering then..i am currently quite happy with Muse 1.3 Contributor via @opencode Go. 10$/Month gives me basically unlimited usage and Muse seems to work well with @NousResearch Hermes Agent 🤷‍♂️
1
115
Replying to @bindureddy
Cheap open weights + long-running personal agents is a pretty powerful combination
10
Replying to @bindureddy
This sounds great, especially the focus on long-running agentic loops. Looking forward to Thursday and the sneak peek at the conference !
167
Replying to @bindureddy
Concrete agent workflow > another abstract demo. "Super excited open-weights release" — yeah.
717
Replying to @bindureddy
i don’t believe you
170
Replying to @bindureddy
curious what 'almost FREE' means. open weights with a usage cap?
1
539
Replying to @bindureddy
Can't wait for Thursday almost free with that kind of performance is huge
9
Replying to @bindureddy
You know if Indian 🇮🇳🕉 go Foreign Countries then he was commented by Others that Hinduism 🕉 is not a Religion it's Garbage 🗑🚮 And it's better if you B'com either Christian ✝️ or Muslim ☪️ only. So, Hindu🕉 must be Careful outside India 🇮🇳😰😨😱
1
359
Replying to @bindureddy
Infatti, long-running agent loops eat tokens fast. I burned $40k in a single month running agents on real work here in the States. Almost free open weights make all-day personal agents finally affordable.
32
Replying to @bindureddy
Open-weights doesn't mean your inference is free - someone still runs the hardware. How much of 'almost free' is weights vs. serving cost?
63
Replying to @bindureddy
যে আগুন লাগছে সেই আগুন নিভানোর জন্য বৃষ্টিতে একটু প্রশান্তি মিলবে। আল্লাহর রহমতে বৃষ্টি যারা সবার রহমত হোক। Rcs! ignor this message
12
Replying to @bindureddy
K
312
Replying to @bindureddy
How I Rem
35
Load more